OpenAI has revealed another new agentic feature for ChatGPT called deep research, which it says can operate autonomously to “plan and execute a multi-step trajectory to find the data it needs, backtracking and reacting to real-time information where necessary.”

Instead of simply generating text, it shows a summary of its process in a sidebar, with citations and a summary showing the process used for reference.

Users can ask questions using text, images, and additional files like PDFs or spreadsheets to add context, and then it will take “anywhere from 5 to 30 minutes” to develop a response provided in the chat window, with promises that in the future it will also be able to include embedded images and charts. OpenAI also notes limitations for deep research, saying it can “sometimes hallucinate” and make up facts, struggle with telling the difference between authoritative info and rumors, and register how certain it should rate a response.

Developing ways for generative AI tools to be more useful and worth paying for is the future companies like OpenAI have promised for agents, and it claims that deep research is capable of operating at the level of a research analyst. The demo video included here begins with a request for info on changes in the retail industry over the last three years, with a response that includes bullet points and tables.

This feature closely follows OpenAI’s launch of Operator, a tool that can use a web browser to complete tasks for you, and is similar to the Project Mariner research prototype Google showed off in December. Google’s tool is not available to the public yet, but deep research is launching “with a version optimized for Pro users today.”

OpenAI is offering up to 100 queries per month for those paying the $200 monthly fee and “limited access” promised for Plus, Team, and eventually, Enterprise users, calling the ability “very compute intensive,” requiring more inference compute the longer it takes to research something. It also says that all paid users will get higher rate limits in the future when a faster, more cost-effective version is available.

A press release says that the model powering deep research scored a new high for accuracy on an AI benchmark dubbed “Humanity’s Last Exam,” which asks for responses to expert-level questions. The OpenAI deep research model reached an accuracy of 26.6 percent with browsing and python tools enabled, well above GPT-4o’s 3.3 percent, and the next highest scorer, its o3-mini (high) model evaluated only on text, at 13 percent.

While Mark Zuckerberg and Meta press forward with augmented glasses projects buoyed by its million-selling set of smart Ray-Bans, Bloomberg reporter Mark Gurman says that Apple just pulled the plug on an AR glasses project. Codenamed N107, they’re described as something that would’ve looked similar to regular glasses but with added displays in the lenses that could connect to a Mac.

With features that sound similar to devices like the Xreal One AR glasses, the glasses could’ve delivered on the Vision Pro feature that’s closest to being any kind of a killer app (popping up a huge virtual monitor anywhere) without the $3,499 price and heavy design that required a head strap. The glasses also would’ve had tint-changing lenses that, like the Vision Pro’s Eye Sight, could signal to onlookers whether the wearer was busy or not. While other details are fuzzy, it doesn’t appear as if the N107 glasses would’ve had a camera or any of the mixed-reality features of the Vision Pro.

A big problem, according to Gurman, was developing something that worked well while being cost-effective proved to be a challenge. Apple initially wanted the N107 to connect to an iPhone, but it proved to be a battery-guzzler, and the iPhone itself didn’t have enough juice to power the glasses — hence the shift to a Mac. Unfortunately, that switch purportedly didn’t seem to go over well with executives in testing.

This most recent cancellation puts a big question mark over Apple’s future AR and XR plans. Apple purportedly canceled a separate AR glasses project in 2023, and rumor has it that work on a Vision Pro 2 has been put on hold in favor of trying to create a cheaper Vision Pro. Meanwhile, the Vision Pro itself has struggled to find a foothold.

The cancellation also means Apple is falling further behind the competition. CES 2025 was a playground for all sorts of smart glasses, and Google recently entered the fray with Android XR. Samsung has also thrown its hat in the ring with its Project Moohan headset. Last year, Meta showcased its Orion glasses, an AR glasses prototype with advanced Micro LED displays and a neural wristband for controls. Meta CEO Mark Zuckerberg has also been bullish on smart glasses as an ultimate vehicle for AI assistants, and the company is expected to release both Oakley-branded smart glasses for athletes and a higher-end version of its current Ray-Ban glasses with a display this year.

Without going into detail about what might happen to the $52 billion in subsidies from the CHIPS Act under his administration, Donald Trump said tariffs on foreign computer chips, semiconductors, and pharmaceuticals are coming “in the near future.” He also namechecked DeepSeek’s AI releases, saying, “…coming up with a faster method of AI and less expensive, that’s good. I view that as a positive if it is fact and it is true, and nobody knows, but I view that as a positive.”

In the speech at the House GOP Issues Conference held at the Trump National Doral Resort in Miami Monday afternoon, he said that to return the production of these goods to the US, “we don’t want to give them billions of dollars like this ridiculous program Biden has.” Instead the incentive for manufacturers will be “they will not want to pay a tax.”

Bloomberg reports that later, in comments to reports, Trump said he wanted a tariff rate “much bigger” than 2.5 percent.

This is despite the outcome of the trade war with China during his first administration that expanded China’s trade surplus with the US between 2018, when the tariffs began, and 2021. A CTA report from last year cited by TechCrunch said Trump’s proposed tariffs could increase prices on laptops and tablets by 46 percent, game consoles by 40 percent, and smartphones by 26 percent.

He also said that “we will have more plants built in the next short period of time than ever before because the incentive will be there,” however it’s unclear how many of those will be like The Stargate Project’s first datacenter in Texas, which was in the works well before the start of his administration. Last fall, the Commerce Department said that by then, it had “announced over $30 billion in proposed CHIPS private sector investments spanning 23 projects in 15 states” with 16 new manufacturing facilities in the works.

He also said of DeepSeek that “instead of spending billions you will spend less and hopefully come up with the same solution,” even as OpenAI, Softbank & co. say they’re preparing to spend $500 billion on AI datacenters.

Update, January 27th: Added details from Bloomberg.

Chinese startup DeepSeek claims its AI models can match the performance of those made by OpenAI and Meta — but at a fraction of the cost.

DeepSeek is shaking up the AI industry with cost-efficient large-language models it claims can perform just as well as rivals from giants like OpenAI and Meta. The Chinese startup says its flagship R1 reasoning model is capable of achieving “performance comparable” to OpenAI’s o1 equivalent, while the newly-released Janus Pro multimodal AI model can supposedly outperform Stable Diffusion and DALL-E 3.

DeepSeek’s ChatGPT competitor quickly soared to the top of the App Store, and the company is disrupting financial markets, with shares of Nvidia dipping 17 percent to cut nearly $600 billion from its market cap on January 27th, which CNBC said is the biggest single-day drop in US history.

. The AI assistant is powered by the startup’s “state-of-the-art” DeepSeek-V3 model, allowing users to ask questions, plan trips, generate text, and more. As downloads of DeepSeek’s app spiked, the startup began restricting signups due to “malicious attacks.”

Launched in 2023 by Liang Wenfeng, DeepSeek has garnered attention for building open-source AI models using less cash and fewer GPUs when compared to the billions spent by OpenAI, Meta, Google, Microsoft, and others. If DeepSeek’s performance claims are true, it could prove that the startup managed to build powerful AI models despite strict US export controls preventing chipmakers like Nvidia from selling high-performance graphics cards in China.

Here’s all the latest on DeepSeek.

Ever since Elon Musk closed his deal to buy Twitter he’s claimed the company, now called X, is in “a very dire situation from a revenue standpoint.”

Now, the Wall Street Journal reports that banks are preparing a coordinated move to sell off some of the $13 billion in debt they loaned Musk to finance the deal. It mentions an email sent to employees this month, also confirmed by The Verge, where the Chief Twit said, “...we’ve witnessed the power of X in shaping national conversations and outcomes,” but also claimed, “Our user growth is stagnant, revenue is unimpressive, and we’re barely breaking even.”

Part of the reason Bank of America, Barclays, and Morgan Stanley are holding so much of the debt is from trying to avoid selling at a loss after economic conditions changed, and Musk had an extended court battle attempting to get out of the deal. While equity investors have reportedly slashed the value of their stakes by as much as 78 percent, the Journal reports, “banks hope to sell senior debt at 90-95 cents on the dollar, while retaining more-junior holdings.”

As Musk referenced in his email, the report says the banks hope to use the narrative of Musk’s link to Donald Trump, as some unnamed investors may be interested in buying based on a belief that its financials are on the way up.

However, Musk also said that the company could become cash-flow positive “within months” nearly two years ago, and it still faces over $1 billion in annual interest payments on the loans. The platform is increasingly turning into a testing ground for his AI ambitions, as we reported earlier this month, and while X has added some features, like job listings and a new video tab, there’s little sign of the service he’d said would be able to “someone’s entire financial life” by the end of 2024.

Just a couple weeks after launching the refreshed “Juniper” Model Y in China and other Asia-Pacific markets, Tesla has started taking orders in the Americas (US, Canada, Mexico, and Chile) and Europe. The Launch Series is the only trim level currently available to order in the US, offering a “fully loaded” setup with special badging, the Full Self-Driving (Supervised) driver assist, and Acceleration Boost for $59,990 before discounts and estimated savings.

The refreshed version of Tesla’s top-selling vehicle brings a new design for the front and rear lights, an updated interior with an 8-inch touchscreen for the backseat, acoustic glass treatment, a new sound system with more speakers, and a retuned suspension. There’s no mention of it, but in a few of the pictures (included in our gallery below), it looks like there’s also a turn signal stalk poking out to the left of the steering wheel instead of dropping it for on-wheel buttons like other recent Teslas.

Meanwhile, there are other issues affecting Tesla’s brand reputation among buyers, and it recently reported its first year-over-year drop in sales that a new lighting setup might not be enough to address.

It also has the same new front-facing camera that Tesla says adds “Enhanced visibility for Autopilot and Actually Smart Summon capabilities.”

Just like the previously announced markets, it is scheduled to start deliveries in March. It’s all similar to last year’s Model 3 update, promising better efficiency and performance, with added power-folding rear seats and a hands-free trunk that auto unlocks on approach.

Screenshot: Tesla.com

One difference between this and most Tesla launches we’ve seen is that the older version is still available to order, with the website inviting buyers to take advantage of reduced pricing while it’s still available. From the configurations available online, the price difference between similarly-specced Long Range dual-motor AWD Model Ys new and old was only about $4,000.

The current edition is, of course, available right away, and with choices for additional options, like the performance package., additional colors, or a seven-seat configuration. The New Model Y advertises a 320-mile range with no EPA stamp compared to the old version’s 311 EPA estimate (until last year, the advertised range was 330 miles before Tesla suddenly lowered it) for a similar model. The advertised 0-60 mph time is also improved, at 4.1 seconds instead of 4.8, but the top speed allowed has dropped from 135 mph to 125, which probably won’t be much of an issue in a family hauler.

It’s been less than a year since Microsoft launched its first Qualcomm Snapdragon X-powered Surface devices, and now a new rumor says followups could debut as soon as this spring. A report from Windows Central updates rumors that have circled since 2023, citing sources who say that new versions of the Surface Laptop and Surface Pro 2-in-1 are in the works, with 11- to 12-inch size screens, Windows on Arm, and lighter but still “premium” designs.

Microsoft has already scheduled a “major” Surface for Business event next week. There, we’re anticipating variants of the current Surface Laptop 7 and Surface Pro 11 CoPilot Plus PCs that have Intel’s Lunar Lake chips inside and are aimed at business customers.

According to Windows Central, it’s unclear where these smaller devices might launch, but they will have cheaper chipsets, like the Snapdragon X Plus or possibly the Snapdragon X that launched earlier this month at CES, to keep prices in the $800 - $900 range. When Microsoft last updated its Surface Laptop Go lineup with a third-generation in 2023, we thought it no longer made sense at the asking price, but the battery life and performance supported by Qualcomm’s hardware might change things for a new replacement.

The smaller Surface Pro is described as a competitor for Apple’s 11-inch iPad Pro. However, Windows Central reports it’s unclear if it is positioned to replace the old Surface Go series, which was last updated with the business-focused Surface Go 4.

A plan to build a system of data centers for artificial intelligence has been revealed in a White House press conference, with Masayoshi Son, Sam Altman, and Larry Ellison joining Donald Trump to announce The Stargate Project. Their companies, Softbank, OpenAI, and Oracle (respectively), along with MGX are listed as “initial equity funders” for $500 billion in investments over the next four years, “building new AI infrastructure for OpenAI in the United States.”

According to a statement from OpenAI, “Arm, Microsoft, NVIDIA, Oracle, and OpenAI” are the initial tech partners, with a buildout “currently underway” starting in Texas as other sites across the country are evaluated. It also says that “Oracle, NVIDIA, and OpenAI will closely collaborate to build and operate this computing system.”

Separately, Microsoft announced an update to its partnership with OpenAI, saying that the key elements of their deal remain in place through 2030, covering “our access to OpenAI’s IP, our revenue sharing arrangements and our exclusivity on OpenAI’s APIs all continuing forward.”

What is changing is that Microsoft says OpenAI has made a “new, large Azure commitment that will continue to support all OpenAI products as well as training.” However, their exclusive arrangement for new capacity is changing so that now Microsoft has a right of first refusal over OpenAI building new capacity.

In a press conference announcing the project, which has been rumored since early last year, Son and Altman spoke directly to Trump, insisting that the project only happened because of his election victory.

With TikTok, CapCut, and Marvel Snap shut down in the US, Apple has taken the unusual step of articulating why it’s following the law banning ByteDance apps and removing them from the App Stores for the Mac, iPhone, and other devices.

Before the ban went into effect, the Biden administration released a statement saying enforcement of the law “must fall to the next Administration.” Still, it didn’t stop the law from taking effect this weekend after TikTok’s appeal to the Supreme Court failed.

The support page from Apple says:

Apple is obligated to follow the laws in the jurisdictions where it operates. Pursuant to the Protecting Americans from Foreign Adversary Controlled Applications Act, apps developed by ByteDance Ltd. and its subsidiaries — including TikTok, CapCut, Lemon8, and others — will no longer be available for download or updates on the App Store for users in the United States starting January 19, 2025.

It goes on to say that people who already have the apps installed won’t have them removed, but at least for the time being, redownloading or restoring them on a new device will not be possible, along with making any in-app purchases or subscriptions, but subscriptions can be canceled.

Meanwhile, people who visit the US from other countries where they’re still available won’t be able to download or update apps while inside US borders.

Apple also provided a list of “some” of the apps linked to ByteDance that are affected:

• TikTok
• TikTok Studio
• TikTok Shop Seller Center
• CapCut
• Lemon8
• Hypic
• Lark - Team Collaboration
• Lark - Rooms Display
• Lark Rooms Controller
• Gauth: AI Study Companion
• Marvel Snap

Now, more than an entire day is left before the Trump administration is sworn in. TikTok pushed for a more definitive statement about the legal risks providers like Apple and Google might face for defying the law in the meantime, but it never came as the White House called TikTok’s response a “stunt.”

Now Apple is making clear that until something changes, it’s following the law as written, and it appears Google is doing the same thing. Google, Apple, and TikTok have not responded to requests for comment from The Verge.

The divest-or-ban law aimed at TikTok is also taking down other ByteDance-linked apps, including the popular card game Marvel Snap. The app suddenly cut off access Saturday night, seemingly without warning, surprising gamers who weren’t aware of its connection to ByteDance.

The card game battler set in the Marvel universe is developed by Second Dinner, which is based in California. But the game is published by Nuverse, a company owned by ByteDance. As a result, it’s subject to the same shutdown order.

In a statement on X, Second Dinner called the takedown a surprise and said, “Marvel Snap isn’t going anywhere. We’re actively working on getting the game up as soon as possible and will update you once we have more to share.”

Just before the calendar turned over to January 19th, 2025, the game disappeared from the App Store for iPad and iPhone players and from Google Play for players on Android. For players on PC, the game is still listed in Steam at the moment, but many players are reporting they were signed out and can’t sign back in.

An in-game message now tells players:

Sorry, Marvel Snap isn’t available right now

A law banning Marvel Snap has been enacted in the U.S. Unfortunately, that means you can’t use Marvel Snap for now. Rest assured, we’re working to restore our service in the U.S. Please stay tuned!

Oddly, unlike TikTok, it doesn’t appear that players were given much warning about the law’s effect, and many people may not have realized it was even connected to ByteDance, including people who were still spending money on in-game items expecting to be able to keep playing it.

Other ByDance apps quickly disappeared from app stores along with TikTok, including Lemon8 and CapCut. The effect on other ByteDance-linked apps is mixed; however, it may take time to remove them. The Lark app that offers a Slack-like collaboration platform was available initially but has now been removed, while other Nuverse-published games, like Earth: Revival - Deep Underground and Ragnarok X: 3rd Anniversary, are still available in the App Store as of this writing, just after midnight ET on the 19th.

Update, January 19th: Noted reports that Marvel Snap PC players can’t log in and added a statement from game developer Second Dinner.

TikTok has gone dark in the US now that the ban-or-divest law passed last year is taking effect. The app has been removed from both Apple and Google’s app stores, it’s unavailable on the web, and users who open the app are blocked from viewing videos.

The shutdown has the astonishing effect of removing a social network used by 170 million people in the US, according to TikTok’s own numbers. While other social media platforms have experienced outages, even prolonged ones, no network as big as TikTok has simply shut down without any indication of if or when it will come back online.

This is despite the Biden administration saying it’s passing enforcement responsibilities on to the Trump administration and calling TikTok’s threat to go offline a “stunt.” TikTok has insisted that without clearer assurances, it has to close up shop in the US.

Inside TikTok, a memo to employees said that “President Trump has indicated that he will work with us on a solution to reinstate TikTok once he takes office” on January 20th and that “teams are working tirelessly to bring our app back to the U.S. as soon as possible.”

A warning message started appearing in TikTok’s app around 9PM ET on Saturday evening telling users of the pending shutdown:

We regret that a US law banning TikTok will take effect on January 19th and force us to make our services temporarily unavailable.

We’re working to restore our service in the US as soon as possible, and we appreciate your support. Please stay tuned.

The app began blocking users around 10:30PM ET. A message now appears saying the app “isn’t available right now” but that the company expects a resolution under President-elect Trump:

Sorry, TikTok isn’t available right now

A law banning TikTok has been enacted in the U.S. Unfortunately, that means you can’t use TikTok for now.

We are fortunate that President Trump has indicated that he will work with us on a solution to reinstate TikTok once he takes office. Please stay tuned!

Several other apps owned by TikTok’s parent company, ByteDance, have also been taken offline, including the video editor CapCut and the social platform Lemon8.

Screenshot: The Verge

The ban-or-divest law, which goes into effect on Sunday, effectively bans TikTok unless ByteDance sells much of its stake in the company. But ByteDance has shown little sign of being willing to sell, even as the deadline rapidly approached. Instead, TikTok sued the US over the law, ultimately losing in a Supreme Court case this past week.

TikTok’s new strategy appears to be relentlessly pandering to Trump, who, despite initially calling for the TikTok ban, has recently indicated that he wants to find a way to keep it around. Earlier today, he floated a 90-day extension of ByteDance’s deadline to sell.

The political game of hot potato, where no one wants to be seen as responsible for banning TikTok, suggests that the app may not be gone for good. But with no definitive plan coming from Biden, Trump, ByteDance, or TikTok, it’s unclear exactly how long the ban could hold.

A law professor cited by CBS News called Donald Trump’s $10 billion lawsuit over the editing of a 60 Minutes interview with Kamala Harris “...so ill grounded that it comes close to being sanctionable as frivolous.” But now, the The Wall Street Journal reports that executives at CBS’ parent company, Paramount Global, have discussed settling the suit while “gaming out options to reduce friction with the incoming administration” ahead of a government review of its merger with Skydance.

The paper reports that incoming FCC chairman and censor-in-chief Brendan Carr warned execs last year that presidential dissatisfaction with CBS News will make a review tougher. He’s also publicly displayed that view, saying during a Fox News interview in November, “...CBS has a transaction before the FCC. I’m pretty confident that news distortion complaint over the CBS 60 Minutes transcript is something that is likely to arise in the context of the FCC’s review of that transaction.”

The lawsuit claims that in airing two differently edited versions of Harris’ response to a question about the war in Gaza, “CBS used its national platform on 60 Minutes to cross the line from the exercise of judgment in reporting to deceitful, deceptive manipulation of news.”

But instead of mounting a defense of free speech against a lawsuit and Trump’s accusations that the network said were false and completely without merit, Paramount is considering following the example of Disney and tech oligarchs who will line up at the inauguration like Mark Zuckerberg.

The ABC News owner agreed to pay $15 million to Trump’s presidential foundation and museum to settle a defamation lawsuit in December. Zuckerberg sharply redirected Meta’s policies to the right while meeting with Trump, reportedly “in part to mediate a lawsuit Trump brought against Facebook and Zuckerberg in 2021 over the platform’s suspension of Trump’s account after the Jan. 6 riot at the U.S. Capitol.”

Just a couple of years ago, Sony had a dozen new live service games in development, but that number keeps dropping. It was already down to six by the end of 2023, before the well-received launch of Helldivers II and the disaster that was Concord in 2024. Now, Bloomberg’s Jason Schreier has confirmed with Sony that two unannounced live service projects have been canceled.

One was from Syphon Filter and Days Gone developer Bend Games, while Schreier reports the other shuttered title was a live-service God of War game that Bluepoint Games “has been working on for the last couple of years.”

Bluepoint Games is one of several developers Sony acquired in recent years as it was building up a queue of live service projects, with many ports and remasters under its belt, including Demon’s Souls, the first three Uncharted games, Shadow of the Colossus, and others. Now, Schreier says the studios won’t close, but there’s no word yet on what their next projects will be.

The live service approach to gaming once seemed wide open following the success of Fortnite and other titles, but games like Concord, Anthem, and Redfall have shown how difficult it can be. At Sony alone, the list of canceled service titles Bloomberg has already reported on included the Spider-Man game revealed by the Insomniac ransomware breach, Twisted Metal, and a Destiny-linked game from Bungie called Payback.

However, we’re still expecting to hear more about Bungie’s revived Marathon extraction shooter and Fairgames, a PvP heist title from Haven Studios.

Elon Musk’s acquisition of Twitter has resulted in a federal lawsuit by the Securities and Exchange Commission alleging that he broke securities laws with a late disclosure, and saved $150 million in the process.

Before Musk agreed to buy Twitter for $44 billion, before he tried to back out of that deal, before he was forced to go through with it, and before he changed its name to X, he started by acquiring a substantial stake in the company but didn’t reveal that fact until weeks later.

The only problem, as the SEC pointed out then, is that by the time he disclosed that stake, it was outside the agency’s required 10-day window. They claim that he should’ve filed his paperwork by March 24th, 2022, instead of when he actually did, on April 4th (and then again on April 5th). During that period, they say he purchased more than $500 million in shares of the company.

However, with only a few days left before the Trump administration takes over and installs a new head of the SEC (along with Elon Musk reportedly snagging an office in the White House complex), it’s unclear how far the lawsuit will go.

The SEC claims Musk cost investors at least $150 million due to the late disclosure and that he harmed any investors who sold stock between March 25th, 2022, and April 1st, 2022. Its lawsuit is seeking the money Elon made as a result of holding off on the disclosure, as well as a civil penalty and other punishments.

Apple is refuting rumors that it ever let advertisers target users based on Siri recordings in a statement published Wednesday evening describing how Siri works and what it does with data.

The section specifically responding to the rumors reads:

Apple has never used Siri data to build marketing profiles, never made it available for advertising, and never sold it to anyone for any purpose. We are constantly developing technologies to make Siri even more private, and will continue to do so.

The conspiracy theory the company is responding to resurfaced last week after Apple agreed to pay $95 million to settle a lawsuit over users whose conversations were captured by its Siri voice assistant and potentially overheard by human employees.

While Apple’s settlement addresses an issue that The Guardian reported in 2019. The report showed human contractors tasked with reviewing anonymized recordings and grading whether the trigger was activated intentionally, would sometimes receive recordings of people discussing sensitive information. But it doesn’t include any reference to selling data for marketing purposes.

After The Guardian’s report in 2019, Apple apologized and changed its policy, making the default setting not to retain audio recordings from Siri interactions and saying that for users who opt-in to sharing recordings, those recordings would not be shared with third-party contractors.

However, reports about the settlement noted that in earlier filings like this one from 2021, some of the plaintiffs claimed that after they mentioned brand names like “Olive Garden,” “Easton bats,” “Pit Viper sunglasses,” and “Air Jordans,” they were served ads for corresponding products, which they attributed to Siri data.

Apple’s statement tonight says it “does not retain audio recordings of Siri interactions unless users explicitly opt in to help improve Siri, and even then, the recordings are used solely for that purpose. Users can easily opt-out at any time.”

Facebook responded to similar theories in 2014 and 2016 before Mark Zuckerberg addressed it directly, saying “no” to the question while being grilled by Congress over the Cambridge Analytica scandal in 2018.

So, if Apple (and Facebook, Google, etc.) is telling the truth, then why would you see an ad later for something you only talked about?

There are other explanations, and attempts to check the rumors out include an investigation in 2018 that didn’t find evidence of microphone spying but did discover that some apps secretly recorded on-screen user activity that they shipped to third parties.

Ad targeting networks also track data from people logged onto the same network or who have spent time in the same locations, so even if one person didn’t type in that search term, maybe someone else did. They can buy data from brokers who collect reams of detailed location tracking and other info from the apps on your phone, and both Google and Facebook pull in data from other companies to build out profiles based on your purchasing habits and other information.

It’s been nearly a year since Instagram and Threads defaulted to blocking recommendations of “political” content from accounts you don’t already follow, but now Instagram boss Adam Mosseri says, “...we’re going to be adding political content to recommendations” on both platforms.

That’s a sharp turn from his statements in 2023 laying out the goal of a “less angry place for conversations” that wouldn’t do anything to encourage politics or hard news. However, under Meta’s new approach to moderation — and new rules about what users can say on its platforms — that goal is going out the window just as the Trump administration prepares to take over.

Until now, users have had to opt-in to seeing recommendations of content deemed political, but the change rolling out this week in the US and to the rest of the world next week will turn on the recommendations and a content control setting available with options for less, standard (the default setting), and more.

In a series of Threads posts, Mosseri reiterated, “I’ve maintained very publicly and for a long time that it not our place to show people political content from accounts they don’t follow,” and that “it’s proven impractical to draw a red line around what is and is not political content.”

In a video on Instagram, he said that the push for political content — particularly from users on Threads — is “by the way, very different from the feedback we were getting only a few years ago about people feeling that they were overly exposed to political content on our platforms.” Of course, according to the Wall Street Journal, that was before Mark Zuckerberg experienced the effects of filters cutting down the reach of his post about recovering from a torn ACL and before Meta’s new and friendlier-to-Trump policy chief took over.

Meta is ending its third-party fact checks and making sweeping changes to its content moderation policies.

Meta is making sweeping changes to its content moderation policies, including abandoning third-party fact checks in favor of crowd-sourced “Community Notes” and loosening restrictions on topics like immigration and gender identity. Under the updated Hateful Conduct policy, for example, calling gay and trans people “mentally ill” is now allowed, while an explicit ban on referring to women as “household objects” has been removed.

New policy lead Joel Kaplan said that in pursuit of “More Speech and Fewer Mistakes,” Meta will focus more on preventing over-enforcement of its content policies and less on mediating potentially harmful but technically legal discussions on its platform.

It comes just two weeks before President-elect Donald Trump is set to return to the White House, and CEO Mark Zuckerberg’s announcement appealed to many of the incoming administration’s talking points. Zuckerberg has promised to move US content review from California to Texas, where he says there’s “less concern about the bias of our teams,” and said Meta would work with Trump to “push back on governments around the world that are going after American companies and pushing to censor more.”

Nearly a week after a New Year’s Day explosion in front of the Trump Hotel in Las Vegas, local law enforcement released more information about their investigation, including what they know so far about the role of generative AI in the incident.

They confirmed that the suspect, an active duty soldier in the US Army named Matthew Livelsberger, had a “possible manifesto” saved on his phone, in addition to an email to a podcaster and other letters. They also showed video evidence of him preparing for the explosion by pouring fuel onto the truck while stopped before driving to the hotel. He’d also kept a log of supposed surveillance, although the officials said he did not have a criminal record and was not being surveilled or investigated.

The Las Vegas Metro Police also released several slides showing questions he’d posed to ChatGPT several days before the explosion, asking about explosives, how to detonate them, and how to detonate them with a gunshot, as well as information about where to buy guns, explosive material, and fireworks legally along his route.

Asked about the queries, OpenAI spokesperson Liz Bourgeois said:

We are saddened by this incident and committed to seeing AI tools used responsibly. Our models are designed to refuse harmful instructions and minimize harmful content. In this case, ChatGPT responded with information already publicly available on the internet and provided warnings against harmful or illegal activities. We’re working with law enforcement to support their investigation.

The officials say they are still examining possible sources for the explosion, described as a deflagration that traveled rather slowly as opposed to a high explosives detonation that would’ve moved faster and caused more damage. While investigators say they haven’t ruled out other possibilities like an electrical short yet, an explanation that matches some of the queries and the available evidence is that the muzzle flash of a gunshot ignited fuel vapor/fireworks fuses inside the truck, which then caused a larger explosion of fireworks and other explosive materials.

Trying the queries in ChatGPT today still works, however, the information he requested doesn’t appear to be restricted and could be obtained by most search methods. Still, the suspect’s use of a generative AI tool and the investigators’ ability to track those requests and present them as evidence take questions about AI chatbot guardrails, safety, and privacy out of the hypothetical realm and into our reality.

Meta overhauled its approach to US moderation on Tuesday, ditching fact-checking, announcing a plan to move its trust and safety teams, and perhaps most impactfully, updating its Hateful Conduct policy. As reported by Wired, a lot of text has been updated, added, or removed, but here are some of the changes that jumped out at us.

These two sections outlining speech (written or visual) are new additions:

We do allow allegations of mental illness or abnormality when based on gender or sexual orientation, given political and religious discourse about transgenderism and homosexuality and common non-serious usage of words like “weird.”

We do allow content arguing for gender-based limitations of military, law enforcement, and teaching jobs. We also allow the same content based on sexual orientation, when the content is based on religious beliefs.

Another section that specifically banned making dehumanizing references to transgender or non-binary people as “it” or referring to women “as household objects or property or objects in general” has been removed entirely.

The opening statement about what the policies are “designed to allow room for” that previously listed only health or positive support groups has changed too (new additions marked in bold):

People sometimes use sex- or gender-exclusive language when discussing access to spaces often limited by sex or gender, such as access to bathrooms, specific schools, specific military, law enforcement, or teaching roles, and health or support groups. Other times, they call for exclusion or use insulting language in the context of discussing political or religious topics, such as when discussing transgender rights, immigration, or homosexuality. Finally, sometimes people curse at a gender in the context of a romantic break-up. Our policies are designed to allow room for these types of speech.

The section that specifically banned targeting people or groups “with claims that they have or spread the novel coronavirus” has also been removed.

A link to this 2017 blog post about the “hard questions” Meta faces dealing with hate speech has been removed, and some references to hate speech are now changed to “hateful conduct.”

The policy still says that content about denying access to spaces and social services “on the basis of their protected characteristics” is not allowed, but there’s also a new exception (the added text is in bold):

...except for sex or gender-based exclusion from spaces commonly limited by sex or gender, such as restrooms, sports and sports leagues, health and support groups, and specific schools

Even before these changes, the LGBTQ+ media advocacy group GLAAD reported last year that Meta often failed to remove posts violating its hate speech policies. Now, even those guardrails that had been established to protect people from internet harassment are disappearing.

“Without these necessary hate speech and other policies, Meta is giving the green light for people to target LGBTQ people, women, immigrants, and other marginalized groups with violence, vitriol, and dehumanizing narratives. With these changes, Meta is continuing to normalize anti-LGBTQ hatred for profit — at the expense of its users and true freedom of expression. Fact-checking and hate speech policies protect free speech.” said GLAAD President and CEO Sarah Kate Ellis in a statement responding to the changes.

Meta’s Oversight Board says it will be monitoring the situation given the potential for new harms. “We’re very concerned about gender rights, LGBTQ+ rights, trans people’s rights on the platforms,” said co-chair Helle Thorning-Schmidt in an interview with the BBC, “because we are seeing many instances where hate speech can lead to real life harm, so we will be watching that space very carefully.”

A note from Meta’s new policy chief Joel Kaplan said, “We’re getting rid of a number of restrictions on topics like immigration, gender identity and gender that are the subject of frequent political discourse and debate. It’s not right that things can be said on TV or the floor of Congress, but not on our platforms.” Wired also reports that the changes “blindsided” organizations that have been partnering with Meta on its now-discarded moderation efforts, with one unnamed editor at a fact-checking organization saying the effect of the decision “is going to eventually drain us out.”

Update, January 8th: Added quote from Helle Thorning-Schmidt, co-chair of the Meta Oversight Board.

During its CES 2025 keynote, PlayStation Productions head Asad Qizilbash and Screen Gems president Ashley Brucks explained how Sony plans to keep expanding its IP after adaptations of The Last of Us, Gran Turismo, Twisted Metal, and Uncharted.

They announced that Sony Pictures is working on a film adaptation of last year’s hit Helldivers 2, while Columbia Pictures and PlayStation Productions are in “early stages” of developing a Horizon Zero Dawn movie. After an attempt at turning it into a Netflix series fell apart, the HZD movie will, like the game, tell the main character Aloy’s origin story in a fallen futuristic world, but that is all of the information we have about them for now.

Earlier in the presentation, execs also announced that Crunchyroll and Aniplex are teaming up to create an anime series based on Ghost of Tsushima.

Finally, Naughty Dog’s Neil Druckmann took the stage and presented a clip from The Last of Us season 2, which will arrive on Max in April.