OpenAI pledges to pay for data center power as Democrats float nationwide moratorium after midterms
OpenAI moves to ease concerns over soaring electricity costs and local impacts of AI data centers as Democrats push for tougher restrictions following sweeping midterm victories.
Coverage for this event has ended.
Anthropic uses AI to accelerate model development amid calls to slow down
Anthropic is using artificial intelligence to accelerate development of its own AI models, even as CEO Dario Amodei has warned about the risks of advancing the technology too quickly, according to research reported Thursday by The Associated Press.
Researchers at the nonprofit Model Evaluation and Threat Research (METR) found that Anthropic, the company behind the Claude chatbot, is improving its AI systems on certain measurable tasks at roughly twice its previous pace.
The findings, based on an 11-week review of internal data and employee interviews, suggest Anthropic's growing reliance on AI to help build more advanced models is driving much of the acceleration, rather than increases in computing power alone.
The rapid progress highlights a challenge facing AI developers: increasingly capable systems could help create their successors faster than companies can evaluate potential safety risks.
Researchers cautioned that their estimates carry substantial uncertainty and may not apply to more complex tasks.
AI stocks sink as Nasdaq enters correction on data center fears
Artificial intelligence stocks suffered steep losses Thursday, pushing the Nasdaq into correction territory as investors worried that tougher restrictions on data center construction could threaten the industry's rapid expansion.
The Nasdaq Composite fell 5.4% in late trading, putting it more than 10% below its recent record, while the S&P 500 dropped 4.3% and the Dow Jones Industrial Average shed 1,844 points, according to CNBC.
AI chipmakers Nvidia and Broadcom each tumbled roughly 9%, while Oracle lost 7% and cloud computing company CoreWeave declined 11%.
The selloff followed sweeping Democratic midterm victories that raised concerns about potential limits on new data centers, adding uncertainty to the billions of dollars technology companies are investing in AI infrastructure.
OpenAI faces lawsuit alleging ChatGPT helped woman plan to poison husband
A North Carolina man is suing OpenAI, alleging ChatGPT helped his estranged wife plan to poison him and conceal her actions, according to a lawsuit reported by USA TODAY.
James Gates alleges the chatbot provided his wife, Cheryl Gates, with information about dangerous substances and avoiding detection. He says he became seriously ill in July 2025 and later discovered handwritten notes outlining a poisoning plan.
Investigators obtained ChatGPT conversations that Gates' attorneys argue show the chatbot helped develop the alleged plot.
Cheryl Gates was indicted in October 2025, but prosecutors dropped all charges in April, citing insufficient evidence that an attempted poisoning occurred. She faces no remaining criminal charges.
OpenAI told USA TODAY it banned the user involved and provided information to law enforcement during the investigation.
Google taps Henry Cavill to show off AI game-building tools with charity challenge
Google is turning to Hollywood star Henry Cavill to showcase how artificial intelligence can help create video games, launching a celebrity-driven experiment that also raises money for men's health, according to GamesRadar+.
The former Superman actor helped develop a retro-style game called Mustache Protocol using Google's AI tools, demonstrating how the technology can turn ideas and prompts into playable experiences without traditional game development.
The project doubles as a fundraising campaign for Movember, with Google pledging $11 for each completed playthrough, up to $500,000.
The game also draws on a memorable moment in Cavill's Hollywood career: the digital removal of his mustache during reshoots for the 2017 superhero film "Justice League."
The promotion comes as technology companies increasingly market AI tools as a way for people without specialized programming experience to create software, games and other digital content.
AI stock selloff sends Nasdaq toward correction as data center fears rattle Wall Street
A sharp selloff in artificial intelligence stocks sent Wall Street tumbling Thursday, pushing the tech-heavy Nasdaq toward correction territory as investors grew increasingly concerned about soaring AI spending and potential restrictions on data center construction.
The Nasdaq plunged nearly 5%, while the S&P 500 sank 3.2% and the Dow Jones Industrial Average shed more than 1,000 points in late-morning trading.
The selloff was fueled in part by concerns that sweeping Democratic victories in Tuesday's midterm elections could lead to tougher restrictions on new AI data centers, threatening the rapid infrastructure expansion powering the technology boom.
Chipmakers were among the hardest hit, with Nvidia falling 6.3% and Advanced Micro Devices tumbling 13.1%. AI cloud computing company CoreWeave dropped 10%.
The Nasdaq fell more than 10% below its Oct. 6 record closing high, putting the index on track to confirm its second correction of the year.
Investors have also grown increasingly skeptical about whether the billions of dollars technology companies are pouring into AI chips, data centers and other infrastructure will generate sufficient returns.
The Associated Press and Reuters contributed to this report.
Meta welcomes OpenAI's data center pledge, says it made same commitment 'long ago'
Meta is welcoming rival OpenAI's pledge to cover the energy costs associated with its massive AI data centers — while pointing out that the Facebook parent company says it made a similar commitment earlier this year.
"This is a good thing," Meta communications official Andy Stone wrote Thursday on X in response to OpenAI's announcement.
"We're glad OpenAI is joining Meta in a long-ago commitment Meta made to pay our own energy and water costs so they're not passed onto consumers, fund new and upgraded grid infrastructure, and support local communities," Stone added.
Meta publicly outlined its commitments in March, pledging to cover the energy and water costs associated with its data centers, invest in new and upgraded power grid infrastructure, and help ensure those expenses are not passed on to local consumers.
The remarks come as OpenAI seeks to ease concerns about the impact of its expanding data center footprint on electricity bills and local communities.
OpenAI also announced new transparency measures Thursday, including a commitment to stop using non-disclosure agreements with public officials during negotiations over future data center sites and to provide residents with information about construction plans and potential traffic, noise and lighting disruptions.
OpenAI's Altman rebuts 2024 Amodei quote, says people should control future
OpenAI CEO Sam Altman is pushing back against a vision of the future outlined by rival Anthropic CEO Dario Amodei, warning that humans must never surrender control to superintelligent artificial intelligence.
"People should always be in control of the future," Altman wrote in a post on X Tuesday evening.
Altman said OpenAI opposes the idea of humans being relegated to activities that "AI systems think makes sense to reward in humans," borrowing language directly from a 2024 essay by Amodei titled "Machines of Loving Grace."
In the essay, Amodei explored how economies might function if AI systems eventually became capable of performing most human work. One possibility he raised involved AI-run economies distributing resources to people based on activities the systems deemed worth rewarding.
Amodei presented the arrangement as one of several speculative possibilities, acknowledging that such systems could create serious problems and potentially dystopian outcomes.
Altman did not mention Amodei by name, but his use of the identical phrase points to the Anthropic chief's earlier writing.
"People should not cede control of the future to superintelligence, and that requirement should ultimately constrain AI progress," Altman added.
Meta acquires startup helping developers manage 'armies' of AI coding agents: report
Meta is making a new play in the artificial intelligence arms race, acquiring a startup that allows software developers to deploy and manage multiple AI coding agents at the same time.
The Facebook and Instagram parent company has acquired Conductor, a startup founded in 2024 that developed a platform for coordinating AI agents working on software projects simultaneously, Axios reported, citing sources familiar with the deal.
The acquisition could help Meta develop its next generation of AI coding tools as tech giants compete to attract software engineers and capture a larger share of the growing market for AI-powered development.
Conductor is led by co-founders Charlie Holtz and Jackson de Campos. Financial terms of the deal were not disclosed.
The move comes as rival OpenAI also expands its developer tools business. On Wednesday, OpenAI told Axios it plans to acquire Pierre Computer, a startup focused on storing and managing AI-generated code.
While Conductor specializes in coordinating multiple AI agents, Pierre focuses on managing the code those agents produce.
Apple slashes employee AI spending limits as tech giant looks to rein in soaring costs: report
Apple is tightening the purse strings on artificial intelligence — even as the tech giant pushes employees to embrace the technology across its operations.
The iPhone maker quietly reduced daily AI spending limits for some engineers from $300 to $200 last month and is considering further cuts that could bring those allowances down to $100, Bloomberg reported Wednesday, citing people familiar with the changes.
Some employees outside engineering could see their daily AI budgets slashed to as little as $25, although workers who need higher limits would be able to request exemptions from senior leadership.
The restrictions target spending on AI "tokens," the units of data processed by models such as OpenAI's ChatGPT, Anthropic's Claude and Google's Gemini. More complex tasks, including writing and debugging software, can burn through large numbers of tokens and drive up costs.
The belt-tightening comes as Apple's AI expenses continue to climb. The company reported a 32% increase in research and development spending in fiscal 2026, driven partly by higher AI and infrastructure costs, according to the outlet.
The changes have reportedly sparked mixed reactions internally, with some employees viewing the previous allowances as excessive while others worry the cuts could signal broader restructuring or potential layoffs.
Apple is not alone in scrutinizing AI spending. Microsoft, Tesla, Uber and Meta have also introduced restrictions or budgets aimed at controlling employee use of AI tools, according to the report.
Apple declined to comment to Bloomberg on the changes.
OpenAI unveils new data center pledges as Democrats push for tougher restrictions
OpenAI is promising greater transparency and community involvement in its massive data center expansion as Democrats push to restrict new facilities following sweeping victories in Tuesday's midterm elections.
The ChatGPT maker announced Thursday that it will stop using non-disclosure agreements with public officials during negotiations over future data center sites, while giving local residents access to construction plans and information about potential traffic, noise and lighting disruptions.
The new policies, which OpenAI says have been in development since August, are aimed at building public trust as communities across the country raise concerns about the economic and environmental effects of the artificial intelligence boom.
"We need to do a better job of really communicating to people, to local communities, about why data centers can be beneficial to them," OpenAI President Greg Brockman said Wednesday.
The announcement comes after Democrats flipped governorships in Texas, Georgia and Ohio, with winning candidates pledging to pause new data center permits and tighten construction regulations.
Of the 41 data center-related ballot measures decided this week, all but one went against data centers, according to Ballotpedia.
OpenAI is also emphasizing the potential economic benefits of its projects. Chris Lehane, the company's chief global affairs officer, said one planned facility in southern Ohio is expected to support 2,500 long-term jobs with an average annual salary of $80,000.
Lehane argued that AI data centers would generate ongoing employment because their computing equipment must be replaced or upgraded every three to five years.
The company faces growing pressure to convince skeptical communities that the rapid expansion of AI infrastructure will deliver lasting local benefits rather than simply increasing demand for electricity and other resources.
Reuters contributed to this report.
Live Coverage begins here