On Wednesday, OpenAI announced the release of two new models—o3 and o4-mini—that combine simulated reasoning capabilities with access to functions like web browsing and coding. These models mark the first time OpenAI's reasoning-focused models can use every ChatGPT tool simultaneously, including visual analysis and image generation.
OpenAI announced o3 in December, and until now, only less capable derivative models named "o3-mini" and "03-mini-high" have been available. However, the new models replace their predecessors—o1 and o3-mini.
OpenAI is rolling out access today for ChatGPT Plus, Pro, and Team users, with Enterprise and Edu customers gaining access next week. Free users can try o4-mini by selecting the "Think" option before submitting queries. OpenAI CEO Sam Altman tweeted that "we expect to release o3-pro to the pro tier in a few weeks."
Should the city of Houston, which proudly bills itself as "Space City," have a prized Space Shuttle orbiter on public display?
More than a decade ago, arguably, the answer was yes. After all, the Space Shuttle program was managed from Johnson Space Center, in southeastern Houston. All the astronauts who flew on the shuttle trained there. And the vehicle was operated out of Mission Control at the Houston-based facility.
But when the final decisions were being made to distribute the shuttles 15 years ago, the Houston community dragged its feet on putting together a competitive proposal. There were also questions about the ability of Space Center Houston to raise funding to house the shuttle within a new display area, which magnified concerns that the historical vehicle, like a Saturn V rocket before it, would be left outside in the region's humid environment. Finally, other cities offered better proposals for displaying the shuttles to the public.
The Trump administration plans to kill the free tax filing program operated by the Internal Revenue Service, the Associated Press reported today, citing two anonymous sources.
The IRS launched Direct File in a pilot for the 2024 tax filing season. It was available to taxpayers in 12 states last year, and was available in 25 states this year. The program's website says the filing tool will be open until October 15 for people who obtained deadline extensions, but it hasn't been updated to account for the plan to end Direct File.
"The program had been in limbo since the start of the Trump administration as Elon Musk and the Department of Government Efficiency have slashed their way through the federal government," the AP article said. "Musk posted in February on his social media site, X, that he had 'deleted' 18F, a government agency that worked on technology projects such as Direct File."
After X (formerly Twitter) announced it would be killing its "Support" account, disgruntled users quickly roasted the social media platform for providing "essentially non-existent" support.
"We'll soon be closing this account to streamline how users can contact us for help," X's Support account posted, explaining that now, paid "subscribers can get support via @Premium, and everyone can get help through our Help Center."
On X, the Support account was one of the few paths that users had to publicly seek support for help requests the platform seemed to be ignoring. For suspended users, it was viewed as a lifeline. Replies to the account were commonly flooded with users trying to get X to fix reported issues, and several seemingly paying users cracked jokes in response to the news that the account would soon be removed.
LG TVs will soon leverage an AI model built for showing advertisements that more closely align with viewers' personal beliefs and emotions. The company plans to incorporate a partner company’s AI tech into its TV software in order to interpret psychological factors impacting a viewer, such as personal interests, personality traits, and lifestyle choices. The aim is to show LG webOS users ads that will emotionally impact them.
The upcoming advertising approach comes via a multi-year licensing deal with Zenapse, a company describing itself as a software-as-a-service marketing platform that can drive advertiser sales “with AI-powered emotional intelligence.” LG will use Zenapse’s technology to divide webOS users into hyper-specific market segments that are supposed to be more informative to advertisers. LG Ad Solutions, LG’s advertising business, announced the partnership on Tuesday.
The technology will be used to inform ads shown on LG smart TVs’ homescreens, free ad-supported TV (FAST) channels, and elsewhere throughout webOS, per StreamTV Insider. LG will also use Zenapse's tech to “expand new software development and go-to-market products," it said. LG didn’t specify the duration of its licensing deal with Zenapse.
On April 2, 2025, our president announced major new taxes on imports from foreign countries (“tariffs”), ranging from 10 percent to 49 percent. The stated goal is to bring manufacturing back to the United States and to “make America wealthy again.”
These tariffs will not work. In fact, they may even do the opposite, fail to bring manufacturing back, and make America poorer in the process.
This article gives the 14 reasons why this is the case, how the United States could bring manufacturing back if it were serious about doing so, and what will ultimately happen with this wrongheaded policy.
A government whistleblower told lawmakers that DOGE's access to National Labor Relations Board (NLRB) systems went far beyond what was needed to analyze agency operations and apparently led to a data breach. NLRB employee Daniel Berulis, a DevSecOps architect, also says he received a threat when he was preparing his whistleblower disclosure.
"Mr. Berulis is coming forward today because of his concern that recent activity by members of the Department of Government Efficiency ('DOGE') have resulted in a significant cybersecurity breach that likely has and continues to expose our government to foreign intelligence and our nation's adversaries," said a letter from the group Whistleblower Aid to the Senate Select Committee on Intelligence leaders and the US Office of Special Counsel.
The letter, Berulis' sworn declaration, and an exhibit with screenshots of technical data are available here. "This declaration details DOGE activity within NLRB, the exfiltration of data from NLRB systems, and—concerningly—near real-time access by users in Russia," Whistleblower Aid Chief Legal Counsel Andrew Bakaj wrote. "Notably, within minutes of DOGE personnel creating user accounts in NLRB systems, on multiple occasions someone or something within Russia attempted to login using all of the valid credentials (e.g. Usernames/Passwords). This, combined with verifiable data being systematically exfiltrated to unknown servers within the continental United States—and perhaps abroad—merits investigation."
The RoboBee lands on a leaf. Credit: Harvard Microrobotics Laboratory
Several years ago, Harvard University roboticist Robert Wood made headlines when his lab constructed RoboBee, a tiny robot capable of partially untethered flight. Over the years, RoboBee has learned to fly, dive, and hover. The latest improvement: RoboBee has learned how to stick the landing, thanks to biomechanical improvements to its landing gear modeled on the crane fly, which has a similar wingspan and body size to the RoboBee platform. The details of this achievement appear in a new paper published in the journal Science Robotics.
As previously reported, the ultimate goal of the RoboBee initiative is to build a swarm of tiny interconnected robots capable of sustained, untethered flight—a significant technological challenge, given the insect-sized scale, which changes the various forces at play. In 2019, Wood's group announced its achievement of the lightest insect-scale robot so far to have achieved sustained, untethered flight—an improved version called the RoboBee X-Wing. In 2021, Wood's group turned its attention to the biomechanics of the mantis shrimp's knock-out punch and built a tiny robot to mimic that movement.
But RoboBee was not forgotten, with the team focusing this time around on achieving more robust landings. “Previously, if we were to go in for a landing, we’d turn off the vehicle a little bit above the ground and just drop it, and pray that it will land upright and safely,” said co-author Christian Chan, one of Wood's graduate students. The trick is to minimize velocity when approaching a surface and then quickly dissipating impact energy. Even something as small and light as RoboBee can generate significant impact energy. The crane fly has long, jointed appendages that enable them to dampen their landings, so the insect served as a useful model for RoboBee's new landing gear.
MIAMI—A decade after its first visit to the state, Formula E returned to Florida this past weekend. The even has come a long way since that first chaotic Miami ePrix: The cars are properly fast now, the racing is both entertaining and quite technical, and at least the trackside advertising banners were in place before the start of the event this time.
It's not the same track, of course. Nor is it anywhere near the Hard Rock Stadium that Formula 1 now fills with ersatz marinas and high-priced hospitality packages during its visit to the area. Despite what the b-roll helicopter shots might have led viewers to believe, we were actually an hour south of the city at a mid-sized oval track next to a landfill in Homestead. Usually, a place that hosts NASCAR races, for Formula E, there was a 2.2-mile (3.5 km) layout that used the straights and infield but not the banked corners.
Formula E has begun to branch out from its original diet of racing exclusively on temporary city center street tracks, having visited Portland International Raceway in Oregon in 2023 and 2024. Despite the bucolic charm of PIR, with its easy bicycle and light rail access, enthusiastic crowd of attendees, and exciting racing, it was only a temporary patch for Formula E. The vast majority of Formula E's fans live outside the US, and Portland means nothing to them, but they've heard of Miami, I was told last year.
In now-rarified comments from experts at the Centers for Disease Control and Prevention, an agency official on Tuesday evening said the explosive measles outbreak mushrooming out of West Texas will require "significant financial resources" to control and that the agency is already struggling to keep up.
"We are scrapping to find the resources and personnel needed to provide support to Texas and other jurisdictions," said David Sugerman, the CDC's lead on its measles team. The agency has been devastated by brutal cuts to CDC staff and funding, including a clawback of more than $11 billion in public health funds that largely went to state health departments.
Sugerman noted that the response to measles outbreaks is generally expensive. "The estimates are that each measles case can be $30,000 to $50,000 for public health response work—and that adds up quite quickly." The costs go to various responses, including on-the-ground response teams, vaccine doses and vaccination clinics, case reporting, contact tracing, mitigation plans, infection prevention, data systems, and other technical assistance to state health departments.
Google may have finally found an application of large language models (LLMs) that even AI skeptics can get behind. The company just released its 2024 Ads Safety report, confirming that it used a collection of newly upgraded AI models to scan for bad ads. The result is a huge increase in suspended spammer and scammer accounts, with fewer malicious ads in front of your eyeballs.
While stressing that it was not asleep at the switch in past years, Google reports that it deployed more than 50 enhanced LLMs to help enforce its ad policy in 2024. Some 97 percent of Google's advertising enforcement involved these AI models, which reportedly require even less data to make a determination. Therefore, it's feasible to tackle rapidly evolving scam tactics.
Google says that its efforts in 2024 resulted in 39.2 million US ad accounts being suspended for fraudulent activities. That's over three times more than the number of suspended accounts in 2023 (12.7 million). The factors that trigger a suspension usually include ad network abuse, improper use of personalization data, false medical claims, trademark infringement, or a mix of violations.
The semiconductor industry is bracing to potentially lose more than $1 billion once Donald Trump announces chip tariffs.
Two sources familiar with discussions between chipmakers and lawmakers last week told Reuters that Applied Materials, Lam Research, and KLA—three of the largest US chip equipment makers—could each lose about "$350 million over a year related to the tariffs." That adds up to likely more than $1 billion in losses between the three, and smaller firms will likely face similarly spiked costs, estimating losses in the tens of millions.
Some chipmakers are already feeling the pain of Trump's trade war, despite a 90-day pause on reciprocal tariffs and a tenuous exception for semiconductors and other electronics.
The Common Vulnerability and Exposures, or CVE, repository holds the answers to some of information security's most vital questions. Namely, which security issue are we talking about, exactly, and how does it work?
The 25-year-old CVE program, an essential part of global cybersecurity, is cited in nearly any discussion or response to a computer security issue, including Ars posts. CVE was at real risk of closure after its contract was set to expire on April 16. The nonprofit MITRE runs CVE and related programs (like Common Weakness Enumeration, or CWE) on a contract with the US Department of Homeland Security (DHS). A letter to CVE board members sent Tuesday by Yosry Barsoum, vice president of MITRE, gave notice of the potential halt to operations.
"If a break in service were to occur, we anticipate multiple impacts to CVE, including deterioration of national vulnerability databases and advisories, tool vendors, incident response operations, and all manner of critical infrastructure," Barsoum wrote.
A New Mexico man is facing federal charges for two separate incidents of alleged arson—one at an Albuquerque Tesla showroom and one at the New Mexico Republican Party’s office—according to a Monday press release from the Department of Justice.
Jamison Wagner, 40, was charged with allegedly setting fire to a building or vehicle used in interstate commerce. The charge can apply to goods manufactured and sold in different states and the facilities that house them—like the Tesla showroom or the Republican office, which also sells MAGA merchandise. DOJ spokesperson Shannon Shevlin tells WIRED that Wagner’s arrest happened on Saturday.
“Let this be the final lesson to those taking part in this ongoing wave of political violence,” Attorney General Pam Bondi said in the Monday press release. “We will arrest you, we will prosecute you, and we will not negotiate. Crimes have consequences.”
In the AI world, a vulnerability called a "prompt injection" has haunted developers since chatbots went mainstream in 2022. Despite numerous attempts to solve this fundamental vulnerability—the digital equivalent of whispering secret instructions to override a system's intended behavior—no one has found a reliable solution. Until now, perhaps.
Google DeepMind has unveiled CaMeL (CApabilities for MachinE Learning), a new approach to stopping prompt-injection attacks that abandons the failed strategy of having AI models police themselves. Instead, CaMeL treats language models as fundamentally untrusted components within a secure software framework, creating clear boundaries between user commands and potentially malicious content.
The new paper grounds CaMeL's design in established software security principles like Control Flow Integrity (CFI), Access Control, and Information Flow Control (IFC), adapting decades of security engineering wisdom to the challenges of LLMs.
There is a signal, born in the earliest days of the cosmos. It’s weak. It’s faint. It can barely register on even the most sensitive of instruments. But it contains a wealth of information about the formation of the first stars, the first galaxies, and the mysteries of the origins of the largest structures in the Universe.
Despite decades of searching for this signal, astronomers have yet to find it. The problem is that our Earth is too noisy, making it nearly impossible to capture this whisper. The solution is to go to the far side of the Moon, using its bulk to shield our sensitive instruments from the cacophony of our planet.
Building telescopes on the far side of the Moon would be the greatest astronomical challenge ever considered by humanity. And it would be worth it.
The rate of autism in a group of 8-year-olds in the US rose from 2.76 percent (1 in 36) in 2020 to 3.22 percent (1 in 31) in 2022, according to a study out Tuesday in the Morbidity and Mortality Weekly Report, a journal published by the Centers for Disease Control and Prevention.
The report's authors—researchers at the CDC and academic institutions across the country—suggest that the slight uptick is likely due to improved access to evaluations in underserved groups, including Black, Hispanic, and low-income communities.
The data comes from the CDC-funded Autism and Developmental Disabilities Monitoring (ADDM) Network. The national network has been tracking the prevalence of autism spectrum disorder (ASD) in 8-year-olds at a handful of sites since 2000, publishing estimates every two years. In 2000, ASD prevalence was 1 in 150, with white children from high-income communities having the highest rates of the developmental disability. In 2020, when the rate hit 1 in 36, it was the first year in which higher ASD rates were seen in underserved communities. That year, researchers also noted that the link between ASD and socioeconomic status evaporated in most of the network.
Google has announced that yet another AI model is coming to Gemini, but this time, it's more than a chatbot. The company's Veo 2 video generator is rolling out to the Gemini app and website, giving paying customers a chance to create short video clips with Google's allegedly state-of-the-art video model.
Veo 2 works like other video generators, including OpenAI's Sora—you input text describing the video you want, and a Google data center churns through tokens until it has an animation. Google claims that Veo 2 was designed to have a solid grasp of real-world physics, particularly the way humans move. Google's examples do look good, but presumably that's why they were chosen.
Prompt: Aerial shot of a grassy cliff onto a sandy beach where waves crash against the shore, a prominent sea stack rises from the ocean near the beach, bathed in the warm, golden light of either sunrise or sunset, capturing the serene beauty of the Pacific coastline.
The Trump White House is proposing to eliminate most federal funding for National Public Radio (NPR) and the Public Broadcasting Service (PBS) and issued a statement yesterday alleging that NPR and PBS "spread radical, woke propaganda disguised as 'news.'"
"The NPR, PBS grift has ripped us off for too long," the White House statement said.
White House budget director Russ Vought drafted a memo for a rescission plan that would eliminate funding already approved by Congress, according to multiple news reports. This includes $1.1 billion for the Corporation for Public Broadcasting (CPB), or about two years' worth of funding for the nonprofit group that provides money to public broadcasting stations.
More than 45 million people in the US are fans of bowling, with national competitions awarding millions of dollars. Bowlers usually rely on instinct and experience, earned through lots and lots of practice, to boost their strike percentage. A team of physicists has come up with a mathematical model to better predict ball trajectories, outlined in a new paper published in the journal AIP Advances. The resulting equations take into account such factors as the composition and resulting pattern of the oil used on bowling lanes, as well as the inevitable asymmetries of bowling balls and player variability.
The authors already had a strong interest in bowling. Three are regular bowlers and quite skilled at the sport; a fourth, Curtis Hooper of Longborough University in the UK, is a coach for Team England at the European Youth Championships. Hooper has been studying the physics of bowling for several years, including an analysis of the 2017 Weber Cup, as well as papers devising mathematical models for the application of lane conditioners and oil patterns in bowling.
The calculations involved in such research are very complicated because there are so many variables that can affect a ball's trajectory after being thrown. Case in point: the thin layer of oil that is applied to bowling lanes, which Hooper found can vary widely in volume and shape among different venues, plus the lack of uniformity in applying the layer, which creates an uneven friction surface.
