• OpenAI, which is owned by a non-profit, wants to adopt a more traditional corporate structure.
• Elon Musk, Meta, charities, academics, and AI entrepreneurs have raised concerns about the plan.
• They question OpenAI's legal process, and whether its non-profit will be compensated fairly.

Sam Altman is attempting a thorny maneuver, transforming OpenAI from a subsidiary of a convoluted non-profit into a more conventional business.

Jungwon Byun knows what that's like.

In 2023, the organization that Byun cofounded, an AI lab named Ought, went through a similar transition. Launched in 2018, Ought built AI research tools, and was structured as a non-profit because its cofounders didn't expect AI to be commercially viable any time soon, Byun said in an interview. Then ChatGPT launched, AI mania took hold, and the leadership team realized their structure might be a hindrance to raising capital and pursuing their mission.

The decision was made to spin out Ought's core product, Elicit, as a for-profit company, along with most of its staff. Byun and several other executives slated to lead the independent Elicit recused themselves from discussions, and a firm specializing in asset valuations was hired to independently appraise the value of Elicit, so the remaining non-profit, Ought, could be paid appropriately. (Byun declined to share the exact price, though she said Elicit chose to pay a premium on top of the recommended price.)

OpenAI's planned transformation is on a far grander scale, but faces similar questions and potentially more gnarly challenges.

Altman wants to extricate OpenAI's revenue-generating business from its non-profit parent, creating a more traditional company that can take in investments and give stock to shareholders who expect big returns.

The process is complicated by OpenAI's complex corporate structure, the billions of dollars it has already raised from investors, and rival Elon Musk's attempts to block the deal.

It's not just Elon though. There's a growing chorus of critics, including entrepreneurs, other companies and charities, investors, academics, and activists, who say OpenAI may be about to make a grave mistake.

Byun is one of them.

"Artificial general intelligence is the most transformative technology of our lifetime and OpenAI is absolutely at the frontier and one of the most important players in that," she told BI. "So I think giving up governance rights to controlling the most important technology of our lifetime is an insanely huge decision and will affect all of humanity at a very great scale for a long time."

Her advice to OpenAI's charitable parent organization: Do not sell to anyone.

"I don't know, obviously, the details of what pressures they're under and what the details are of the options they're considering, and how it affects the mission," Byun said. "But from what I can see, I absolutely would not sell those governance rights or give them up." OpenAI didn't respond to requests for comment over the past week.

A convoluted structure

OpenAI was established as a non-profit in 2015. Over time, its need for capital to pay for AI development prompted it to establish a for-profit subsidiary to raise venture funding and pursue business partnerships and revenue-generating products. This (and various other subsidiaries) are controlled by the nonprofit, which is controlled by a board of directors chaired by Bret Taylor, alongside Altman, Fidji Simo, Larry Summers, and other business luminaries.

The board's main legal obligation is to further the non-profit's charitable mission, which is to build "safe and beneficial artificial general intelligence for the benefit of humanity," rather than delivering returns to shareholders.

OpenAI now seeks to transition to a more conventional corporate structure, turning the for-profit subsidiary into an independent public benefit company that controls "OpenAI's operations and business."

PBCs are relatively common. This approach takes into account the interests of shareholders, other stakeholders, and the public. With investors taken care of, OpenAI has said this will help the startup raise "necessary capital with conventional terms like others in this space."

Finding fair value

The remaining non-profit will receive shares in the public benefit company "at a fair valuation determined by independent financial advisors."

That may be the ultimate challenge — and one made more complex by potential conflicts of interest.

OpenAI investors and others focused on the startup's business-focused future may prefer to pay the non-profit less for its assets and control, leaving the new for-profit entity with more financial firepower. However, some of these people could be in positions to influence the valuation and decision-making process.

Elon Musk, an OpenAI cofounder who acrimoniously split from Altman, has sued the startup over its plans. Musk and a group of other investors also made a bid in February to acquire the nonprofit parent OpenAI Inc. for $97.4 billion. That was more than half of the total value of the entire startup, based on its most-recent funding round.

"It is in the public's interest to ensure that OpenAI, Inc. is compensated at fair market value," the Musk consortium said in a statement. "That value cannot be determined by insiders negotiating on both sides of the same table."

OpenAI's board has an obligation to make sure the nonprofit is getting a fair value for giving up its assets and control, according to Ellen Aprill, an expert on charities at the UCLA School of Law.

"One of the very big issues is what is fair market value?" she told BI. "In particular, how much should the nonprofit get for giving up control?

"Part of what Musk and his group were trying to do with this offer is to set at least some floor, some sense of what the fair market value would be," she added. "Putting it out for auction — let's see what people would pay."

Should Altman recuse himself?

Taylor, the OpenAI board chairman, officially rebuffed Musk. "OpenAI is not for sale, and the board has unanimously rejected Mr. Musk's latest attempt to disrupt his competition," Taylor wrote in a statement. "Any potential reorganization of OpenAI will strengthen our nonprofit and its mission to ensure AGI benefits all of humanity."

But days before that, Altman publicly shot the bid down with a post on X that mocked the struggles of the Musk-owned social network: "no thank you but we will buy twitter for $9.74 billion if you want."

Altman's post prompted some discussion of whether he was, or should be, recused from this process.

"When it comes to looking at a conflicted transaction like this one, it's much easier to satisfy the court's review if you take people who are on both sides of their transaction out of the negotiations," said Peter Molk, a law professor at the University of Florida.

However, he noted that under Delaware law, Altman isn't legally required to remove himself. Molk also cautioned against recusing too early.

"There's value in his being there. Along the way, he can provide useful insight and information that's actually useful for OpenAI, for the nonprofit, if they're going to go ahead with that process," Molk explained. "You want to try and strike a balance that allows them to do their job, that helps a transaction occur in a way that's actually beneficial to all parties, but then have them take themselves out of the picture once you're beyond that point."

Meta and others pile on

The existing for-profit parts of OpenAI have already raised billions of dollars in pursuit of "Artificial General Intelligence," or AGI, which the startup defines as systems that outperform humans at most economically valuable work. This has raised hackles among some consumer advocates.

Public Citizen, a consumer watchdog, has repeatedly assailed the AI startup, accusing it of betraying its nonprofit mission. "The nonprofit board has behaved as a subordinate to the for-profit, and has done nothing that evidences any commitment to the nonprofit mission," said Robert Weissman, the co-president of Public Citizen.

The solution is to forcibly dissolve the nonprofit and auction off its assets for the benefit of a new independent charity not linked to the current OpenAI board, he added.

Facebook parent company Meta raised similar concerns, writing to the California Attorney General Rob Bonta in December to ask him to step in.

"Taking advantage of this non-profit status, OpenAI raised billions of dollars in capital from investors to further its purported mission," the company wrote. "Now, OpenAI wants to change its status while retaining all of the benefits that enabled it to reach the point it has today. That is wrong. OpenAI should not be allowed to flout the law by taking and reappropriating assets it built as a charity and using them for potentially enormous private gains."

In late January, a group of 25 charities, including one started by eBay founder Pierre Omidyar and his wife, piled on.

The coalition wrote to the California AG, urging him "to take prompt legal action to ensure that OpenAI's assets are not illegally diverted for private gain."

Dubious legal justifications

Rose Chan-Loui, a law professor at UCLA specializing in non-profit and philanthropy law, is dubious of the legal justifications for OpenAI's plan.

She said the proposed changes would amount to a change in purpose for OpenAI's current non-profit. To do that under Delaware law, requires one of four criteria to be met:

• Is the charity's original purpose now illegal to fulfill?
• Has the mission become impossible to fulfill?
• Is it now impracticable to fulfill?
• Or too wasteful to fulfill.

"We don't think any of those apply here," Chan-Loui said. "We keep going back to — your purpose is not to be first in the race to AGI. Your purpose is to make sure that AGI is to protect, and ensure safe development of AGI."

The lesson of other nonprofits

OpenAI's existing structure — a non-profit that controls a for-profit subsidiary — is unusual, but by no means unheard of.

The Mozilla Foundation, for example, owns a tech company, Mozilla Corporation, that produces the Firefox internet browser. "Non-profit organizations can not only coexist with commercial enterprises but also challenge them to do better," a Mozilla spokesperson said in a statement. "Non-profit commitments provide the kind of public interest compass we don't see enough of in tech."

Another parallel comes from the healthcare sector. In the 1990s, in the face of mounting financial pressure, the Blue Cross Blue Shield Association allowed its affiliated nonprofit plan organizations to transition to for-profit businesses. Many subsequently did so, to the chagrin of some.

"The mission of the non-profit in most Blue Cross Blue Shield plans was to provide affordable, accessible health care to people," a Consumers Union attorney told NPR in 2010. "And now the mission is to make money for stockholders."

The most substantial threat to OpenAI's plans though, might not be Elon Musk or other critics. The California and Delaware Attorneys General are both making inquiries into the specifics of the transition.

"The current beneficiaries of OpenAI have an interest in ensuring that charitable assets are not transferred to private interests without due consideration," Delaware AG Kathleen Jennings wrote.

AGs have meaningful legal powers to get involved. In 2002, Pennsylvania's Attorney General intervened in a planned deal by the non-profit foundation that controls the for-profit Hershey chocolate brand to sell its assets to Wrigley for $12 billion, citing the potential harm to the town of Hershey.

The sale was ultimately called off.

• Suchir Balaji, an ex-OpenAI researcher who raised concerns about his employer, died in November.
• His parents have demanded a thorough investigation into his death.
• An autopsy report said authorities found a gun registered in Balaji's name.

San Francisco authorities said a case involving the death of a former OpenAI researcher in November is "closed," citing, among other evidence, a self-inflicted gunshot wound from a pistol registered to the victim, documents shared with Business Insider show.

Suchir Balaji, an ex-OpenAI employee who last year raised concerns about copyright issues regarding his former employer, was found dead in his San Francisco apartment on November 26. He was 26 years old.

Local authorities at the time ruled Balaji's cause of death as an apparent suicide with no evidence of foul play.

Balaji's mother, Poornima Ramarao, previously told BI that the family would seek a private autopsy. Attorneys representing Balaji's parents also filed a petition against the city and county of San Francisco in January, demanding the immediate release of public records related to their son's case.

In a joint letter to the family attorneys dated February 14 from the San Francisco Police Department and the medical examiner's office, authorities wrote that Balaji was found lying alone in his bathroom with no signs of a "forced entry to the single door leading into the residence."

In the police's investigative findings, the letter said, a firearm was found underneath Balaji's leg.

"The pistol found at the scene was purchased and registered to Mr. Balaji in January 2024," the letter said.

Gun residue particles were found on Balaji's right and left hand, the autopsy report reads. The report said Balaji sustained "a severe firearm injury."

"There were no signs of unrelated injuries or bruising to Mr. Balaji," authorities wrote.

Authorities said in the letter to the Balaji family's attorneys that they had seen the preliminary report from the independent autopsy the family sought and that the findings were "consistent" with the medical examiner's office's findings, including the entrance and trajectory of the bullet.

"The SFPD found no evidence or information to establish that Mr. Balaji died of means other than a suicide by self-inflicted gunshot wound to the head," the letter from SFPD and medical examiner's office said. "If you learn of any new information or evidence, please provide it to SFPD. SFPD considers this case closed and will only reopen the criminal investigation if there is a basis for a chargeable offense and the statute of limitations has not expired."

Kevin Rooney, an attorney representing Balaji's parents, told BI the family had received the letter and the autopsy report. He did not have further comment.

Balaji was an OpenAI researcher for four years, helping the AI startup gather data for its GPT-4 model. A former colleague told BI in an interview that Balaji was instrumental in developing ChatGPT's early predecessor.

"He was one of the true geniuses at OpenAI," Tarun Gogineni, a research scientist at OpenAI, said.

In October, just a few months after Balaji left the company, The New York Times published an interview in which Balaji accused his former employer of violating copyright law.

OpenAI is involved in several other legal battles concerning copyright violations. On November 18, Balaji was named in the Times' lawsuit as a custodian who could provide supporting legal documents for the case.

An OpenAI spokesperson did not respond to a request for comment.

Balaji's mother, Ramarao, previously told BI she wasn't satisfied with the police investigation. To raise awareness about her son's case, she organized public vigils and made multiple media appearances.

Ramarao wrote in a post on X on Friday afternoon that she wanted a "transparent investigation."

A spokesperson for the medical examiner's office did not address Ramarao's social media post when responding to BI's request for comment. An SFPD spokesperson did not respond to a request for comment.

Three weeks before Christmas, in 2016, Greg Robillard received an email from his lawyer. Attached was a brusque, five-page letter from an attorney representing his former employer, a tech startup named Opal Labs based in Portland, Oregon.

The company intended to sue him over his GitHub account.

The soft-spoken Portland resident had worked in the technology industry since 1999, wrangling databases and writing software for various companies over the years — a nonprofit, a healthcare startup, a marketing firm. He had signed up for GitHub in 2012, as the code management tool was becoming the industry standard, using it like tens of millions of other tech workers do — to collaborate on coding projects with colleagues, or to show a portfolio of work to prospective employers.

When he interviewed for the job at Opal Labs, a downtown marketing-tech firm, he sent an executive his GitHub profile to provide a sample of code he'd written. Once he started as a lead enterprise engineer, he followed his new colleagues' lead and published code to his public profile, using the tool to keep his manager apprised of his progress. When he left the company six months later, Robillard again used GitHub as part of his job hunt — recreating from memory a project he'd worked on at Opal derived from decades-old public code.

Now, Opal's lawyers were saying that code Robillard had posted to the platform amounted to confidential trade secrets. It was the warning shot in what would become a four-year legal nightmare.

Looking back, things soured for Robillard just a month after he joined the company, when he invited some of his new coworkers to his 42nd birthday party.

Afterward, he felt office attitudes shift. As at many tech firms, Opal's workforce skewed young; newly aware of his age, colleagues now called him "old Greg," he later testified, and the CEO mocked his use of Facebook as something "older people" did. He also said he heard a job applicant dismissed as "some old guy in his 40s."

In May 2015, after Robillard missed a scheduled meeting with a client without alerting his colleagues — he said he was looking after his son as his wife was leaving town for a funeral — he was fired.

Opal told Robillard he was terminated for the absence and other performance issues, but Robillard recalled colleagues missing meetings without consequence. He started wondering if he'd crossed an invisible barrier — he'd become old, without even noticing — and had been discriminated against. He hired an employment lawyer and, about a year after his termination, filed suit against Opal alleging age discrimination, among other claims.

Opal's lawyers filed a response denying his allegations, and the parties began discovery. That's when the company fired back with the letter — and a countersuit against Robillard, accusing him of violating state and federal laws protecting trade secrets.

Suddenly he was being cast not as a victim of discrimination, but as an intellectual property thief.

"It was horrifying," Robillard said in an interview. "They were using this pressure, this lever that I hadn't even thought about before. And it felt like a very upside-down experience."

Across the United States, Business Insider has found, companies large and small are wielding trade secrets law as a potent legal strategy against workers who have accused them of wrongdoing.

There's the aerospace employee who was laid off after accusing colleagues of Islamophobic abuse; he filed a discrimination claim and was then countersued over documents he'd emailed himself while on the job. There was the music school director who said he was subjected to sexual harassment and a hostile work environment and was sued after leaving to start a rival business. And there was the healthcare worker who was fired after raising suspicions of Medicare fraud and sued her employer, alleging retaliation, only to be countersued over the documents she had taken to substantiate her claims.

Using trade secret law to fend off worker complaints is so far a niche strategy. In interviews, some attorneys said they'd never seen it deployed. But others have witnessed it multiple times and said its use was on the upswing. Employment law attorneys said its rise has been aided by a recent federal law, a shift to remote work that blurred personal and professional boundaries, and a growing corporate surveillance culture in which a worker's every action can be tracked and weaponized.

Some workers were sued after gathering evidence of perceived wrongdoing in the workplace, what some attorneys call "self-help discovery" — despite whistleblower protections in the law.

In other cases, what workers described as run-of-the-mill professional activities were reframed as trade secret misappropriation: documents printed out to review at home, a work email forwarded to a personal account for safekeeping, a computer shared with a romantic partner, a password sent to a colleague for an urgent task — or code saved to a GitHub profile.

"Tools that were intended to help businesses protect themselves against other businesses and corporate espionage are now used by corporations against their own employees, to protect the company at all costs," said Jennifer Gibson, the cofounder of Psst, a whistleblower-support group.

A beverage company's secret recipe. An innovative manufacturing process. A social media platform's proprietary feed algorithm. A sales team's closely guarded customer list.

Simply put, a trade secret is a valuable, proprietary piece of business information that a company has taken steps to keep under wraps. Allowing businesses to safeguard information, processes, prototypes, code, and other secrets from rivals is foundational to corporate law. Long governed by the states, the law was fully codified at the federal level in 2016, when President Barack Obama signed legislation called the Defend Trade Secrets Act.

The legislation didn't depart radically from existing trade secrets laws in the states, but it created a single, nationwide framework. It also made it far easier for companies to bring trade secrets claims in federal courts, which are perceived by many attorneys as favorable to companies, in part because of stricter rules and faster-moving dockets.

Under both state and federal law, a trade secret has been misappropriated if it was acquired or disclosed without consent or through improper means — whether or not it made its way to a rival firm. "It doesn't matter at all if it went to a competitor," said Michael Risch, a professor of law at Villanova University. "If you obtain information that you're not entitled to have, then potentially you can be liable."

It's now being deployed by employers facing claims of discrimination or wrongdoing.

Attorneys who represent workers in employment lawsuits often agree to take a case on contingency — meaning they bear the legal costs and are paid only if they win or settle. But when it comes to defending a retaliatory counterclaim, many require their clients to pay up front.

Such cases are expensive: Determining whether any particular piece of information is a trade secret is a fact-intensive question that must be decided by a court or arbitrator, an often time-consuming and costly prospect. A survey by the American Intellectual Property Law Association found that in 2022, the median cost of litigating a trade secret case with less than $1 million at stake was $750,000. (Cases that don't make it to trial, the survey found, could still run to tens or hundreds of thousands of dollars.)

"Both potential reputational costs of having that fight happening in court, and a legal bill an ordinary person simply couldn't afford," said Chris Seaman, a professor at Washington and Lee School of Law, mean "the in terrorem effect" — the fear factor — "would be very significant."

In any trade secret case, the odds are strongly against the defendant: A study by the legal analytics firm Lex Machina of all trade secret cases in US District Courts from 2021 to 2023 found that plaintiffs prevailed four times as often as defendants. Mostly, the cases settled. For an employee, a settlement might entail abandoning their discrimination or whistleblower claim, accepting a smaller payout, or even paying damages to their employer.

"Did all these lawyers go to a conference we didn't hear about?" asked Jack Moran, a Cleveland attorney who has represented multiple clients who were accused of trade secret misappropriation after raising their own claims of corporate wrongdoing. "Just one after the other is adopting: 'The best defense is a good offense,' even if the employer doesn't have particularly viable claims."

Business Insider examined dozens of federal trade secrets claims filed by companies over the past decade against current and former employees who also had filed a claim against their employer.

One such suit was filed by the United States Olympic & Paralympic Committee. Despite a dearth of competitors — the committee is the sole entity responsible for overseeing Team USA — the committee filed a trade secrets claim against a former executive after he reported allegations of wrongdoing.

Bill Moreau, a chiropractor, was promoted by the committee to vice president of sports medicine in 2017. Throughout 2018 and 2019, he repeatedly flagged what he said were troubling issues, according to court filings: first, a statutory rape claim involving a 15-year-old Paralympic athlete; and later, what he said were violations of athletes' confidential medical records and failures of care for athletes with suicidal thoughts.

In May 2019, he was fired.

The Olympic committee said Moreau was terminated so the organization could replace him with someone who had a doctor of medicine, rather than a doctor of chiropractic. But Moreau believed he was fired in retaliation for what he described as whistleblower complaints. He sued the following year alleging wrongful discharge.

His approach to document handling soon made him a target.

From time to time, he had emailed himself documents — he said he'd done so to facilitate remote work, as he preferred his personal computer's large screen — or printed them out and took them home. His lawyers said in court filings that this was common practice among the committee's employees.

He said he also saved documents to use in his push for reform at the committee and to protect himself from becoming a "scapegoat." After his termination, he let his attorneys review some of them.

To the Olympic committee's attorneys, his actions were grounds for a countersuit alleging trade secret misappropriation: they accused him of seeking to profit from the documents and said his actions damaged the committee's relationship with athletes. The case ground on for three years before reaching a settlement.

One of Moreau's attorneys, Michael Fairhurst, said he'd noticed that "employers are becoming more aggressive in lodging complaints of misappropriation and the like against workers who have blown the whistle on corruption, discrimination, or other misconduct."

These cases, he said, "can often crush employees into silence."

In the cases BI reviewed, misappropriation allegations weren't just deployed as a defense maneuver after an employee sued. In many instances, employers sued aggrieved employees proactively under trade secrets law — putting workers on the backfoot from the get-go.

Many of the cases hinged on methods of forensic document accounting that didn't exist just a few decades ago. Software can now track every keystroke or file opened during a worker's time with a company; every message and email can be retained for posterity. When someone speaks out against an employer, this entire digital history can be scrutinized for use against them.

Eric Bachman, the founder of Bachman Law, in Bethesda, Maryland, said he had begun checking whether prospective clients — deliberately or inadvertently — retained any files from their employment, to head off this line of attack.

"Generally speaking, you just want to be aware of the issue because we know now that companies are being more aggressive about pursuing these claims" over trade secrets, he said.

Of the various types of legal claims companies can make against workers, misappropriation allegations can be particularly damaging to an employee's professional reputation, Bachman said, because companies will often say employees "stole" proprietary information or accuse them of "theft."

"This is meant to, and does, sound like the employee committed a crime of some sort," Bachman said, "which is a far different implication than something more benign-sounding like breach of contract."

When the Defend Trade Secrets Act was being drafted, its authors were attuned to how it might be abused and attempted a remedy: a whistleblower immunity provision.

This clause sought to protect workers who take documents in an effort to prove that their employer was engaged in illegal activity, offering them immunity from prosecution for what might otherwise constitute misappropriation.

The provision is narrowly defined, applying only to workers who took documents for the sole purpose of sharing them with their attorneys or a government agency. That means employees who seek to expose discrimination or wrongdoing who preserve documents in other ways — by sharing them with the media, say, or seeking guidance from a confidant — may be left unprotected.

"You often have whistleblowers who are having to make decisions in less than ideal circumstances, when they may be emotionally impacted from what's happening," said Gibson, the whistleblower advocate. "Those aren't circumstances that lead to the best decision-making."

In practice, the provision has sometimes failed to protect even the most fastidious would-be whistleblowers.

In one of the earliest cases to test the law, an insurance company, Unum, sued a former employee, Timothy Loftus, alleging misappropriation after he took boxes of documents to investigate what he suspected was illegal activity. The worker invoked the whistleblower exemption, but the judge rejected the motion, saying it wasn't clear that Loftus took the documents solely for the purpose of investigating whether a crime had taken place. (The case was later settled.)

Peter Menell, a UC Berkeley law professor who was a key architect of the exemption, found the court's interpretation outrageous. In a January 2017 research paper, he warned that under this reading of the law, "any trade secret owner can require a whistleblower to defend a trade secret lawsuit merely by alleging that there is a dispute over the employee's motivation for providing trade secret documents to their attorney." Unum was happier with the outcome; Emily Downing-Baer, a company spokesperson, said the decision demonstrated "the value of the Defend Trade Secrets Act."

One Georgia case appeared to be exactly what the whistleblower exemption was designed for.

Christopher Young had suspected his employer, an Alpharetta-headquartered cement company named Argos, was engaged in illegal price-fixing — coordinating with competitors to inflate cement and concrete prices for government projects. He photographed and secreted home extensive paper documents to support his suspicions and used them as the basis for a qui tam lawsuit — a suit brought by a private citizen on behalf of the US government.

Young filed his qui tam case anonymously, but after company officials discovered he was behind it, Argos sued him in 2018, alleging trade secret violations. The company claimed that not all the documents Young took were related to his qui tam lawsuit and that he gave the documents to a competitor. Young's attorney moved to have the case dismissed under the whistleblower clause. The judge denied the motion, citing Loftus, even as she agreed with Young's argument that Argos had provided no plausible evidence that he had disclosed trade secrets to another company.

It turned out Young's whistleblower claims had merit. In 2021 federal prosecutors charged Argos with price-fixing and antitrust behavior, which the company admitted, agreeing to pay a $20 million criminal penalty and cooperate with the ongoing investigation. Argos' case against Young was halted only when the Department of Justice's antitrust division intervened.

Even without whistleblower status, it is illegal for a company to retaliate against someone who engages in protected activity, such as filing a complaint over gender or racial discrimination. But again and again, courts have ruled that trade secrets claims were not tantamount to retaliation unless a defendant could prove they were unreasonable, "baseless," or "frivolous."

In one 2018 case that made it to the 6th Circuit Court of Appeals, an Ohio insurance worker, Steven Aday, was sued over claims of trade secrets violations after he filed a lawsuit accusing a former employer, Westfield Insurance, of age discrimination. At issue was the fact that he'd emailed himself documents; even though he said the company knew other employees had done the same thing, his claim of retaliation was rejected in court. "There are many reasons an employer would not litigate every infraction employees commit," the majority wrote. "However, after an employee has hauled an employer into court, it is entirely reasonable for the employer to file its claims for minor infractions." (Aday ultimately prevailed on the trade secret misappropriation allegations, because his employer couldn't prove he violated company policy.)

Robillard, the Oregon engineer, also accused his former employer Opal of retaliation after Opal's countersuit alleging trade secrets violations. His claim was dismissed, too.

"Even if the court assumes a retaliatory motive, Plaintiff's retaliation claims fail," Magistrate Judge John Acosta wrote, finding that Robillard had not proved the trade secret claims were baseless.

As a partner at Dhillon Law Group and head of the firm's employment law division, John-Paul Deol has sat on both sides of the table. At times, he has defended employees against trade secrets counterclaims, and at times he had filed such claims on behalf of employers himself.

"It's a huge piece of leverage over an employee," the San Francisco attorney said. "On the defense side, it's one of the first things you look at."

On multiple occasions when he's been preparing to file suit over workplace discrimination, a client's employer has threatened a trade secrets countersuit, he said. Sometimes that threat alone has been enough to prompt a worker to abandon their claim. The cost and reputational risk of defending against one of these claims, he said, is just too high.

This dynamic is why it is so difficult to estimate the true impact of trade secrets law on workers, and how much it constrains their ability to speak out about wrongdoing. Multiple attorneys told BI that the mere suggestion of a trade secrets counterclaim had been enough to persuade workers to withdraw or settle their complaints. In addition, many trade secret cases may be litigated entirely in private arbitration, out of sight.

"Even where the employee has a perfectly good defense to that claim, they know it's going to cost them money to fight it out," said Bachman, the Bethesda attorney. "It does factor into their decision about whether or not to bring a case — or to settle it before they ever even have a chance to bring it."

Carl Boja, who'd worked as a salesperson near Akron, Ohio, brought a disability claim against a former employer, a roadside services firm called TA, related to his March 2024 dismissal. In an interview, Moran, the Cleveland attorney, who represented Boja, said TA responded by telling him the company was preparing a trade secrets suit against his client over files Boja had emailed himself. If Boja would drop his claim, Moran recalled, TA would agree not to bring the case. (TA and its attorneys did not respond to requests for comment.)

"He didn't know that he was going to get fired," Moran said. "So it's not as if he took these things and then quit and then formed a competitor." Before Boja had time to consider the offer, Moran said, TA filed suit.

A mandatory arbitration agreement Boja had signed with TA as a condition of employment consigned his disability claim to private arbitration. But TA filed the trade secrets claim against him in open federal court, claiming he took the documents to a competitor — potentially risking Boja's reputation and future job opportunities. It also hurt him financially: Moran had agreed to represent Boja on contingency in the discrimination arbitration, but Boja is now paying him out-of-pocket to defend against the trade secrets claim.

"Dealing with TA's case against me has been emotionally difficult," Boja said. "I have trouble sleeping or focusing on anything else."

Jerry Talton is also battling to save his reputation after being sued over document downloads. In October 2022, while employed as an executive at the fintech firm Carta in New York City, he wrote to the company's board of directors claiming a litany of wrongdoing, including bullying by senior executives, sexually inappropriate comments by colleagues, and employees failing to disclose financial conflicts of interest.

Shortly before the company placed him on leave to investigate, Talton downloaded a raft of internal documents. An accidental email to the company's general counsel prompted a review of his IT activity, which surfaced the downloads. A filing Talton made in court said he'd offered to return the documents through counsel, but the company had declined and fired him, then filed suit claiming trade secret violations. Amanda Taggart, a Carta spokesperson, flatly denied that he'd offered to return the documents. Carta said in court filings that Talton had abused the company's trust and risked revealing business information it had spent years developing.

Court filings show that the litigation process agonizing for Talton. Carta unearthed salacious texts he had sent to sexual partners from company devices and made them public in court filings. "Talton became ridiculed around the world," his attorneys wrote in pursuing his ongoing counterclaims including retaliation, defamation, and wrongful discharge. "A good job offer was withdrawn. He has not been able to get another."

"It's very easy to use these tools — that are there for a good reason — for purposes that are not," Talton said in an interview. He described trade secrets laws as "a blunt weapon" that could be deployed against employees "because you don't like their opinions, or you don't like the manner in which they've spoken up about what they've observed."

Shortly before Thanksgiving in 2020, the chief operating officer of a Harrisburg, Pennsylvania, debt collection firm slapped Nicole Durenleau across the face. She had previously reported inappropriate sexual comments and touching by the COO, Shell Sharma, to an employee relations representative, her attorney wrote in a court filing.

Durenleau left the company, NRA Group, in February 2021 and promptly sent a demand letter complaining of sexual harassment and seeking back pay and compensation. She also reported Sharma to the police for striking her, and he was convicted of harassment that May.

The company counterattacked, suing her on hacking charges in April, and later added a trade secrets claim.

The company's claims hung on a thin thread. About a month before Durenleau quit, her manager requested a login credential for a website that tracks state licenses. Durenleau was out sick that day, so she asked a colleague, Jamie Badaczewski, to log into her account on NRA's computer network to find it. Badaczewski retrieved a spreadsheet of work logins Durenleau had created and sent it along to Durenleau's personal email address so she could share it with her manager. The company's attorneys asserted that creating the spreadsheet and having it sent to her personal email was trade secret misappropriation. The company sued Badaczewski too.

More than two years later, a federal judge ruled that because passwords don't have "independent economic value," they're not trade secrets. Counterclaims by Durenleau and Badaczewski alleging a hostile work environment and workplace retaliation remain ongoing, though Sharma died in early 2024. NRA Group has appealed the dismissal of the trade secrets claim, telling BI that the spreadsheet "contained highly sensitive trade secret information." (Durenleau and Badaczewski's attorney, Cory Iannacone, said his clients had "always maintained that NRA's claims were without merit" and would present their discrimination and retaliation claims at trial.)

Defendants like Durenleau and Badaczewski who prevail when facing trade secrets claims will face headwinds if they seek to recoup their legal costs. The federal trade secrets law sets a high bar for defendants to be awarded fees, requiring the case to be brought in "bad faith."

Though judges sometimes find the evidence of trade secrets violations thin, they rarely find that the claims involved bad faith. In a recent case in California, a federal judge called the case before her "objectively specious," yet didn't approve the award of legal fees.

The result: Even when a company's trade secrets claim fails, it can exact financial harm.

For any trade secret case that makes it all the way to trial, Seaman, the Washington and Lee professor, said, "You're looking down the barrel of a million dollars or more in attorneys fees and expenses, and even if you win, no guarantee that you can get it back."

For Robillard, the Oregon engineer, the legal process was grueling.

His attempts to have the trade secrets claims against him dismissed through summary judgment failed. He sat for depositions and turned over reams of old files for discovery. He was fearful of talking about the case to anyone except his partner, concerned he'd be accused of another violation.

"That isolation is a big factor," he said. "Paranoia is real, because you don't know who's looking or watching."

After he was fired from Opal Labs, he took on other technology jobs but found himself experiencing panic attacks and other symptoms of post-traumatic stress disorder. He left the tech sector in 2015 and began training as a therapist, with a focus on trauma and relationship issues.

The parties finally reached a settlement in March 2021, just a few months before a trial was scheduled to begin. All claims were dismissed with prejudice. Opal agreed to pay a settlement of about $220,000.

"We understand that parties may have different recollections or perspectives on the events in question," Colin Savoy, Opal's chief legal officer, wrote by email. "Opal remains confident in our actions, our defense, and our firm belief that the facts fully supported each and every defense and claim we presented. As a technology company, protecting our intellectual property is both necessary and appropriate."

Of the settlement money, an estimated $94,000 went to pay expert witnesses, according to the office of Robillard's attorney, Courtney Angeli. Around $82,000 went to the law firm — much of which was spent on filing fees, deposition costs, and other expenses, Angeli said.

"I still believe it was 'worth it' because employers who mistreat employees make me crazy," she wrote by email. "But it was not a financially lucrative engagement in the slightest."

After five years of litigation, Robillard ended up with about $44,000, less than half his annual salary at Opal before he was fired.

Rob Price is a senior correspondent for Business Insider and writes features and investigations about the technology industry. His Signal number is +1 650-636-6268, and his email is [email protected].

Narrativa is among a crop of startups seizing on the artificial intelligence boom to enthusiastically automate writing tasks that would once have fallen to humans. From penning regulatory documentation for Pfizer to zhuzhing up marketing copy for insurance and e-commerce firms and helping generate breaking news articles for The Wall Street Journal, the Los Angeles-based Narrativa boasts roughly 50 clients in various industries. But one of its core focus areas, comprising a quarter of its business, is a little more polarizing than the rest: gambling.

Working with major industry players like 888 and Betway, Narrativa uses large language models to pump out everything from automated summaries of sports games to SEO-friendly reviews of online casino games and promotional social media posts. With no humans required, the 20-person company's AI tools produce 10 million words a month for gambling clients — the effective output of 170-odd full-time writers producing a grueling 3,000 words a day. It's all in service of enticing gamblers to place more bets.

"You want to create a community, you want people coming back for more," Matthew Rector, Narrativa's vice president of content, says. "You want to foster that environment, and our content helps facilitate that."

Sam Altman, Elon Musk, Satya Nadella, and the tech industry's other top impresarios talk a big game about how AI may one day attain sentience, solve the climate crisis, and lead society to a post-scarcity economy. Today, though, the technology is being embraced by traditional industries for more prosaic — and mercenary — aims. Key among them is the gambling industry, which is rapidly adopting AI for everything from writing alluring online marketing copy to identifying and helping problem gamblers to tracking people and perfecting the physical layout of casinos.

The ultimate goal: to harvest ever more money from gamblers, by profiling them, feeding them content and games personalized to their whims, and cajoling them to stay longer and make bigger bets.

The gambling industry, much like AI, is in the middle of an unprecedented gold rush. In 2018, a US Supreme Court ruling allowed states to legalize sports betting; nearly 40 states since did exactly that. Major investments have since flooded in, with some gambling stocks hitting record highs and private-equity firms jumping into multibillion-dollar deals with gambling and casino companies.

Meanwhile, consumers' wallets have been emptying: Americans bet a record $120 billion in 2023, according to the American Gaming Association. A study by California researchers released in 2024 estimated that legalized gambling across America may result in as many as 30,000 bankruptcies and an additional $8 billion in debt collections each year. Another paper out of Kansas found that average household investments in the stock market dropped in states where gambling was legalized by roughly $50 per quarter. (In Brazil, which legalized online gambling in 2018, as much as one-fifth of welfare money is now spent directly on gambling, the AP reported in November.)

Under the hood, artificial intelligence is helping power this surge.

Several online betting platforms, for example, offer "micro bets," which allow gamblers to bet in real time throughout the game — who gets the next touchdown or makes the next tackle, whether the next play will be a run or a pass. AI companies like SimpleBet (recently acquired by DraftKings for $195 million) have automated processes that allow the maximum number of possible micro bets to increase by an order of magnitude. No longer bottlenecked by the capabilities of human sportsbook odds calculators, every moment of a sports game can be turned into a wager. Won your bet that Lamar Jackson would throw on 2nd and 10? Why not bet again that he'll scramble for the first down on 3rd and 3?

Physical casinos are also looking to harness AI for efficiency gains. nQube, a Canadian startup run by a physics professor, uses machine learning to optimize the placement of slot machines on casino floors, profiling players and the performance of one-armed bandits individually and collectively — replacing an older generation of floor-manager intuition and basic analysis. Some of nQube's findings have been counterintuitive: It turns out that removing the total number of slot machines can often increase the casino's "win," if the machines are arranged in a way that redirects players to games where they'll make larger bets.

Jason Feige, a cofounder of nQube, had been working on computational astrophysics when his partner, Stasi Baran, found a scientific paper about the problem of optimal casino floor planning. Though neither had any gaming experience, they both realized their work could be a fit. "I like math and I like hard problems, and she brought me just a monster of a problem and I just fell in love with it," Feige says. "I have never seen data as clean and as comprehensive as what you see in this industry, largely because it is so heavily regulated. But that combined with the kind of powerful AI systems that I've been building, it was just such a natural fit. I just absolutely fell in love with the industry."

The prospect of deeper AI integration is definitely in the air. At one of the gambling industry's biggest events, G2E, a glitzy conference held in Las Vegas in September, there were packed panels on AI in sports betting, women in AI, AI-powered behavioral analytics and "responsible gambling," and AI for customer relationships.

One of the most enticing, and controversial, uses of AI in gambling is customizing casinos — virtual and on the floor — to each gambler.

Just as Netflix uses machine learning and data science to tailor each user's feed to what they're most likely to binge, the startup Future Anthem uses similar tools to keep users hooked on casino websites. The UK-based software provider builds a personalized, dynamic homepage, presenting the exact right game — bingo, slots, poker — to cater to a player's desires at the exact right moment, offering bonuses if the player is getting dejected and keeping them betting for longer.

"We have machine-learning models that are understanding and humanizing that player, that player data," says Ian Tibot, Future Anthem's chief commercial officer. "We see every single spin of a slot, we see every single bet, and we actually understand the experience that the player is having by creating the concept of a session out of that data, and that allows us to understand changes in patterns of behavior."

Brick-and-mortar casinos are also digitally profiling their users. Some locations have RFID chips embedded in every gambling chip, tracking how each gambler is playing and building a profile that automatically directs human workers to intervene as needed — an extra free drink here, a bonus spin there. "Before it used to be like a pit boss maybe having their eyes on 40 players across five tables" to monitor bet sizes and manually assign perks and freebies, says Kasra Ghaharian, a researcher at the University of Nevada, Las Vegas, International Gaming Institute. "It wasn't very accurate," he says. AI allows casinos to "be much more precise in how you're tracking that activity."

Beyond using AI for efficiency gains and user profiling, the industry's ultimate vision for employing the technology is much more ambitious — and unsettling.

In a research paper published last May, the consultancy giant Deloitte's Global Lottery and Gambling Centre of Excellence predicted a future where every game could be personalized in real time to appeal to individual gamblers. Generative AI, the authors wrote, could "allow the games themselves to generate content based on the explicit or even implicit actions of players, from instantly generated new items and playing levels to in-game characters that can have lifelike discussions."

The technology, they continued, could create "individually themed online slot games that can respond to a player's voice and even generate novel content in response to a player's behavior and game history." Generative AI chatbots the players could talk to, games with themes automatically tailored to their preference — the ultimate filter bubble. Social media's endlessly personalized carousel of content is already notoriously addictive, and the damaging parasocial relationships that can be formed with AI chatbots are currently under a microscope following reports of suicide and self-harm linked to a popular provider. Adding these elements to the famously powerful money-extraction machine that is online gambling is a potent combination.

Gambling is historically a human-centric business — gamblers try their luck against the house, for better or worse. But Christina Thakor-Rankin, a veteran industry consultant based in the United Kingdom, dreams of an automatically managed brick-and-mortar casino in the years and decades ahead, akin to Amazon's automated Go convenience stores, with unnecessary human staff costs eating into the casino's margins.

"Look at the amount of operating expenditure required in terms of serving customers, monitoring customers, keeping them safe, people who work in the cage or the cashier pit bosses. What if you had a casino that was very similar to the new generation of self-service Amazon stores where you don't need cash and you don't need people?" she asked. "How would that kind of technology transform a world of sportsbooks, but also land-based casinos?"

At least one casino workers union, the Culinary Workers Union, has raised concerns about the risk of blue-collar jobs in Las Vegas being automated. In 2019, The Nevada Independent reported that between 38% and 65% of casino jobs (depending on the study cited) in the south of the state could be automated over the next decade and a half, calling the city "one of the most vulnerable to automation in the entire country."

But for many gamblers, a trip to Vegas isn't just a transaction — it's an experience, punctuated by banter with dealers, table service, shows, and the seedy glamour of the strip. It remains to be seen if they will accept a robot substitute.

Artificial intelligence may be a moneymaker for gambling companies, but the companies say it's also something else: a remedy for problem gambling.

Playtech, a European gambling software provider, is one of several firms using AI to try to suss out when a gambler is demonstrating signs of addiction or problematic play and intervene. Part of this is recognizing a player's patterns — larger-than-usual bets, or declined deposits, or playing at unexpected times — and interjecting with prompts suggesting they take a break. (Future Anthem says its systems can also detect aberrant behavior and automatically check in with target gamblers.)

"Online gambling companies have lots, tons of data about their players because every single bet or every spin on a slot, every single deposit, the time you spend online, there's lots of information," says Francesco Rodano, Playech's chief policy officer. "So we train an AI model to analyze this behavior and recognize possible harmful patterns."

Playtech and other gambling companies are also developing chatbots that gamblers can talk to about addiction — the logic being that because gambling addiction is stigmatized, addicts may actually be more comfortable talking to a nonhuman.

Rodano acknowledges that the same technology that could help problem gamblers could also be used to exacerbate their addictions. "If you use a tool like ours to identify vulnerable players, in theory, you could use that information to target them — which is the opposite of what the tool is intended for, which is totally unethical," he says. "If you operate in a regulated market, it's very unlikely to happen because the regulator would notice that and clamp down."

Amid frothy valuations and wild hype, there's a risk of overstating the technology's near-term promise — particularly the more giddy ideas, such as Thakor-Rankin's predictions of a Caesar's Palace augmented with robotic, voice-activated "Centurions." And some industry insiders say that what's now called AI might be considered statistics or big data, just rebranded.

"I like to say that we've been doing AI since before it was cool and this new age of AI hype — it's been very interesting to navigate because on the one hand, everyone wants to talk about AI, which is great for us," says Stasi Baran, nQube's cofounder.

"On the other hand, there's so much noise to sift through for our customers and for us as well to determine, everyone says that they've got an AI product, but what's actually real and what actually brings real value? I mean, that's difficult to determine. I think there's a lot of overnight AI experts out there, and that concerns us."

This AI frothiness isn't unique to the gambling industry, and the space has long had a nose for innovations that boost its bottom line — from the development of electromechanical slot machines in the 1960s to the creation of loyalty programs for high rollers in the '80s. As ever, the house always wins.

Rob Price is a senior correspondent for Business Insider and writes features and investigations about the technology industry. His Signal number is +1 650-636-6268, and his email is [email protected].

Kevin Weil showed no sign of the heat when he took to the stage at the Marriott Marquis in downtown San Francisco on an unseasonably sweaty early October morning.

Trim and tan, outfitted in the de rigueur Silicon Valley uniform of a slim-fit gray tee, stonewashed blue jeans, and an Apple Watch Ultra, the chief product officer of OpenAI talked easily about the ambitious artificial intelligence-powered future his red-hot employer was building.

"You could imagine a world where you ask it a hard question about how you cure some particular form of cancer, and you let it think for five hours, five days, five months," Weil prophesied in response to a question about the "reasoning capabilities" of AI.

But neither he nor his onstage interlocutor, Anyscale cofounder Robert Nishihara, ever acknowledged the elephant in the room: Weil wasn't supposed to be there.

The event, the AI infrastructure conference Ray Summit, had originally booked OpenAI's chief technology officer, Mira Murati, to speak. Just days before, the high-profile executive had abruptly quit, prompting the last-minute swap. Murati's exit added to a long list of recent departures from OpenAI, one of the world's most valuable and hyped startups, even as it closed a historic $6.6 billion funding round on the day Weil spoke.

If Sam Altman is the starry-eyed visionary of OpenAI, Weil is its executor. He leads a product team that turns blue-sky research into products and services it can sell, putting him at the center of a philosophical rift that has caused spectacular upheaval at the company, which was recently valued at $157 billion.

In two years, OpenAI went from a nonprofit lab nominally working to develop digital intelligence for the public good to a world-famous startup that puts out shiny new products and models every few months. The company is now attempting to become a for-profit operation to lure would-be backers to write bigger checks, which it needs to scale its business. Altman recently announced that the company's flagship product, ChatGPT, now has 300 million weekly users, triple the number it had a year ago.

Along with this astronomical growth, OpenAI has succumbed to a brain drain: Its chief research officer, head of AGI readiness, co-lead of its video-generation model Sora, and the list goes on. While this has sparked alarm bells in some corners of the tech industry, it has also elevated the profile of the senior leaders who have remained. That includes Weil, a relative OpenAI newcomer who joined in June and rapidly became one of its most notable ambassadors.

At a point when employee voices of dissent were growing louder, 41-year-old Weil arrived as a steady-handed product guru with a Midas touch. He was a longtime Twitter insider who created products that made the social media company money during a revolving door of chief executives.

At Instagram, he helped kneecap Snapchat's growth with competitive product releases such as Stories and live video. (Not everything he touched turned to gold, though. Weil also led Facebook's headlong charge into financial services as a cofounder of Libra, its ill-fated stablecoin.)

Weil declined to be interviewed for this article, which is based on conversations with five former senior colleagues, four of whom spoke on the record, and his past public interviews.

Twitter and Facebook were no strangers to chaos and scandal, but even their most challenging times are rivaled by OpenAI's workplace turmoil. There was a failed coup last year, bitter feuds among some workers, and an ongoing existential arms race to build "digital gods."

Weil's peers are certain he's the man to promote harmony. He's spent the better part of 15 years cranking out products that mostly delighted users and made money. He also has something his new employer desperately needs: The ability to pursue the company's best interests and balance human emotions at the same time.

Kevin Weil isn't a household name. But for those in the know in Silicon Valley, he's something better: "He's the get-shit-done guy," said James Everingham, who worked with Weil at Instagram and Facebook.

Sarah Friar, OpenAI's new head of finance, has long admired Weil's product chops. Former Twitter boss Adam Bain called Weil "Twitter's secret weapon" for driving ad revenue. Everingham also described Weil as a workhorse who never shrank from a deadline.

"He brings that stamina, that dogged focus on the outcome," said April Underwood, a venture capitalist who worked closely with Weil on Twitter's ad products.

Weil's relentless work ethic and people skills are recurring themes among former colleagues. At Facebook, two colleagues said he would often badge into the office first and fire off messages late into the night.

One colleague from Planet, a satellite imagery company where Weil worked as president until May of 2024, recalled how he started each week by posting in Slack the top three things on his mind. According to Everingham, Weil's a rare breed of product manager who codes almost as well as he writes memos. That endeared Weil to the engineers he needed to build products.

Weil shrugs off the pressures of work with long runs and Diet Mountain Dew. Every birthday, he runs his age in miles and, in June, marked his 41st year with a 41-mile jaunt near his home in the posh California suburb of Portola Valley, according to a public Instagram post. Weil lives with his venture capitalist wife, Elizabeth, and three children.

"He is a little bit superhuman in just the sheer amount that he works and works out," said Ashley Johnson, chief financial officer and now president of Planet.

Weil set aside a Stanford doctorate in theoretical particle physics to cave out a path in tech, according to a 2017 speech he gave at his alma mater. In 2009, Weil landed at Twitter as a data scientist. At the time, Twitter had little revenue, never mind profit, to show for its many millions of users. That left investors wondering how Twitter would translate its popularity into money.

When Twitter began developing ads a year later, Weil stepped up to lead it. Katie Jacobs Stanton, a venture capitalist who overlapped with Weil at Twitter, said employees debated how to show ads in a way that didn't degrade the user experience, pitting engineers against marketers. Weil threaded the needle. Under his oversight, Twitter launched ads that looked like tweets inside the feed. AdAge reported in 2011 that hundreds of brands had embraced the format, helping to establish the cash flow that made Twitter's IPO possible in 2013.

Then, in early 2016, Instagram cofounder Kevin Systrom asked Weil to dinner as Snapchat was nipping at the Facebook-owned photo app's heels.

Instagram needed to get users, especially teens, to post more; Systrom wanted a pinch-hitter to get new features designed and added to the app and into the hands of Instagram users. Weil told CNBC in 2017 that he had already resigned from Twitter with plans to train for a 50-mile ultramarathon snaking the American River in California's Central Valley. He took the Instagram job and, a month later, finished fifth in the race.

The product leader wasted no time in the photo-sharing battle. In just a few months, Instagram rolled out a feature similar to Snapchat's disappearing photos and videos, then added popular face filters and also introduced a feed-ranking algorithm to highlight more relevant content. According to Instagram, its user base doubled within two years of Weil's arrival, reaching 1 billion monthly users by 2018; Snapchat's earnings statements during the same period indicated that its user growth had flatlined.

Everingham, his former colleague, recounted how Weil identified Stories as Instagram's killer feature and assembled a nimble team of engineers under the CEO's supervision to build it.

"He had this clarity of thinking I haven't seen in anyone else," said Everingham, now an engineering leader of developer infrastructure at Meta.

Weil's tenure at Instagram was defined by a controversial strategy: borrowing liberally from the competition. The features that became Instagram staples had their genesis in a rival's playbook, which neither Systrom nor Weil denied in interviews with Vox and TechCrunch over the years.

This strategy is particularly poignant as Weil steps into his new role at OpenAI, a company navigating a thicket of copyright lawsuits. New outlets, authors, and celebrities have sued OpenAI over using their work to train its large language models.

That's far from the only struggle the product chief faces. He's positioned as one of Altman's top lieutenants at a time when OpenAI's famed brain trust is leaving in droves, often to start competitor companies that may threaten OpenAI's early dominance in the space.

Former OpenAI employees and siblings Dario and Daniela Amodei founded Anthropic, one of OpenAI's most notable competitors, in 2021. Former chief scientist Ilya Sutskever has raised $1 billion for his new venture, Safe Superintelligence. Ex-researcher Aravind Srinivas is working on an AI-powered search engine, Perplexity. In early December, Murati, the former CTO, told Wired she was "figuring out" what her new venture would look like, though it's unclear if her startup will directly compete with OpenAI.

Another threat comes from open-source AI models championed by the likes of Meta. If these free-to-use systems prove good enough for most users, it will make it far harder for OpenAI to effectively monetize its AI models and, ultimately, turn a profit.

Because with over $20 billion funding, according to PitchBook data, and only an estimated $3.7 billion in revenue in 2024 plus losses of $5 billion, according to leaked documents obtained by The New York Times in September), OpenAI still faces a long road to prove that it can deliver a return on the unprecedented volumes of capital plowed into the company. And that's without even getting into the growing concern that improvements in AI models are slowing.

Onstage at the Ray Summit in October, Weil shrugged off the threat of competition. When asked about whether the current gap in quality between open-source models and OpenAI's premium AI products will shrink, he quipped: "I mean, we're certainly going to do our best to make it grow."

In theory, a proven leader like Weil could help guide the operation past the power struggles and talent losses toward a more steady state. By all accounts, he's a deft conciliator who can empathize with the needs and concerns of multiple stakeholders.

"He's somewhat of a diplomat," said Jacobs Stanton, his former Twitter colleague.

At Facebook, where Weil helped develop a stablecoin backed by a basket of international currencies, he had to mediate between crypto natives, the product purists who wanted to take a more user-friendly, less decentralized approach, and policymakers, according to two former Facebook colleagues. The project ultimately couldn't overcome regulatory roadblocks and an exodus of corporate partners; Meta shuttered the project in 2022, selling $182 million worth of assets to Silvergate Bank.

How Weil's experience maps onto his current role remains to be seen. With its several thousand employees, regulators scrutinizing its every move, and 300 million weekly active users of ChatGPT, OpenAI is more complex than any company Weil's stepped into before. In just six months since his arrival, OpenAI has rolled out a large language model that can solve more complex problems via a process the company calls "reasoning," and a search engine within ChatGPT.

Additionally, OpenAI launched a voice mode to talk to ChatGPT, which Weil personally tested as a "universal translator" during recent trips to Seoul and Tokyo. Reflecting on the release, Weil shared on LinkedIn, "It feels normal to me now, but two years ago I wouldn't have believed it was possible."

Last week OpenAI announced an ambitious "12 days of shipmas," a festive product sprint likely to keep Weil working long hours. While he's managed to keep the proverbial plates spinning in his professional life, there's been one noticeable casualty: his workout routine. App data from Strava shows that he's been logging fewer hours of cycling and running each month since he joined OpenAI.

Asked about Weil's exercise, OpenAI spokesperson Niko Felix said Weil recorded 96 minutes of physical activity a day in November. "I would say he's doing quite alright," Felix said.

Melia Russell is a senior correspondent at Business Insider, covering startups and venture capital. Her Signal number is +1 603-913-3085, and her email is [email protected].

Rob Price is a senior correspondent for Business Insider and writes features and investigations about the technology industry. His Signal number is +1 650-636-6268, and his email is [email protected].

• A notable portion of Silicon Valley's electorate has steadily shifted toward Donald Trump.
• While the region remains solidly Democrat, the GOP candidate made gains in this election.
• In Santa Clara, San Mateo, and San Francisco counties, Trump's vote share increased by several points.

Silicon Valley, long considered a progressive stronghold, has begun to shift toward Donald Trump, according to new voting data analyzed by Business Insider.

Across the three San Francisco Bay Area counties that constitute the epicenter of America's tech industry, there was a marked bump in support for Donald Trump in the latest presidential election, and a corresponding decrease in support for Kamala Harris and the Democrats, provisional voter data shows.

California can take weeks to tally all its ballots, so the final totals won't be known for some time. But with well over 95% of the votes counted in San Francisco, Santa Clara, and San Mateo counties as of this week, a trend has emerged.

Among the voting citizens of Santa Clara county, home of Apple, Google, and Nvidia, 28.1% cast their ballot for Donald Trump this year. That's up from four years ago, and notably higher than in 2016.

Meanwhile, 68.1% voted for Harris. That was down from 72.6% who backed Joe Biden four years ago, and 73.1% who voted for Hilary Clinton in 2016, according to data compiled by the California Secretary of State.

In San Mateo county, where Facebook parent company Meta is headquartered, the vote shifted from 77.9%-20.2% Republican-Democrat to 73.5%-23.2% this cycle.

And San Francisco county's vote for Trump went from 9.3% in 2016 to 15.5% this year.

These numbers show that Trump and Republicans are still a long way off from gaining any real majorities in Silicon Valley. However, even this relatively small shift suggests that the region, and the tech industry, is becoming less firmly Democrat — and comes as some tech leaders grow more pragmatic about politics.

Meta CEO Mark Zuckerberg previously publicly supported Democratic candidates, but didn't this year. Venture capital power players Marc Andreessen and Ben Horowitz said they were supporting Trump this summer, though Horowitz also provided funding for the Harris campaign.

While Elon Musk is based in Texas these days, many of his companies, including Tesla, are still very active in Silicon Valley. He sunk hundreds of millions of dollars into helping reelect Trump this year, is undoubtedly the president-elect's most high-profile supporter in tech.

Incoming Vice President JD Vance once worked in venture capital and is seen as a protégé of PayPal cofounder and investor Peter Thiel, one of tech's most influential conservative voices.

Some of this comes down to self-interest. Trump, with Musk in support, has promised to cut regulations, which could help Silicon Valley startups grow more quickly.

"The most exciting thing of all will be putting Elon in charge of government efficiency," said Ben Narasin, a VC based in Atherton, in San Mateo county. "He's going to take a chainsaw through calcified butter, and it's going to be awesome to attack the bloat and overreach that we've had in the government."

The trend toward Trump in the Silicon Valley electorate mirrors statewide and national trends. Nationwide, the Democrats fell from 51.3% to 48.3%, while the GOP rose from 46.9% to 49.9%.

In California, Democrat voteshare dropped from 63.5% in 2020 to 58.6%, as of the data available on Tuesday.

Editor's note: This story was originally published on November 12. It was updated on November 26 after more ballots were counted. This new data did not materially change the story's findings.

Healthcare startup Forward just closed its doors last week, but its executives are already onto their next venture.

Adrian Aoun, Forward's former CEO, is cofounding a new company alongside the startup's former head of operations, Jonathan Lesser, and its former head of product, design, and engineering, Bali Raghavan, according to messages seen by Business Insider.

Lesser told former Forward employees that Aoun will serve as a cofounder, board member and advisor to the new startup, per the messages.

It's unclear what the startup will focus on as a business or if any of Forward's previous investors will be involved in the new venture. Forward raised over $650 million from top investors like Khosla Ventures, Softbank, and Salesforce's Marc Benioff.

Aoun, Lesser, and Raghavan didn't respond to requests for comment from Business Insider.

It's an ultra-quick turnaround from Forward's leaders after the startup announced on November 13 that it would immediately shutter its more than a dozen locations and lay off its nearly 200 employees.

Launched in 2017, Forward sought to reimagine healthcare with ultra-modern clinics and grabbed a $100 million Series E round in November 2023 to power its AI-powered healthcare kiosks, called CarePods.

But a series of logistical and technical challenges — from the CarePods' self-service blood draws not working as expected, to patients getting stuck inside the large metal boxes — hindered Forward's CarePods rollout, Business Insider found. And patients didn't flock to the CarePods as Forward hoped, according to former employees.

On a Friday episode of The Information's More or Less podcast, Aoun said he's already hearing interest from Forward's investors to back his next venture.

"On Tuesday morning, I was like, I've decided, I'm doing nothing for six months to a year. I need a break, I'm licking my wounds, this really f—ing sucks," he said. But that day, he said, one of Forward's biggest investors called him to ask, "So what are you doing next, and can I cut you a term sheet?"

"We just burned not quite half a billion dollars on an idea. And a lot of people's reaction is, what are you doing next? Let's do it again," he continued. "What sort of special ass culture did we create in Silicon Valley where this is reality? This is absurd."

Got a tip on Forward? Contact Rebecca Torrence via encrypted messaging app Signal (+1 423-987-0320), or email [email protected] and contact Rob Price via encrypted messaging app Signal (+1 650-636-6268), email ([email protected] or [email protected]).