GOP senator joins push for AI liability as Google announces new frontier AI model

Synopsys CEO Sassine Ghazi said Thursday that the chip-design software company's partnership with OpenAI would expand if the artificial intelligence giant moves forward with an initial public offering.

"It will expand. Absolutely. It will expand because then their balance sheet will allow them to do more: more chips, more models. They'll be able to support and be able to do more. So which is a great opportunity," Ghazi said during an appearance on FOX Business' "The Claman Countdown."

"Every customer loves optionality," Ghazi said, explaining that companies want to build their own chips to better optimize their software.

Synopsys also announced a multiyear $1 billion agreement with Amazon to help develop custom chips and AWS infrastructure. Ghazi said Synopsys will receive additional royalties as the chips move into production.

President Donald Trump defended the rapid expansion of data centers powering artificial intelligence, arguing the U.S. risks losing the infrastructure and its economic benefits to other countries if it does not continue building them.

"If we don't build them, they'll be built someplace else," Trump told Time Magazine in an interview published Thursday. "I know China is totally over it. China wants every data center you're going to have."

While Trump acknowledged that he would not necessarily want a data center immediately next to his own home, he argued that communities hosting data centers can benefit financially from the facilities and pointed to Loudoun County, Virginia, as an example.

"The data centers. You explain it to the Americans. They make a fortune. They're going to lower your taxes. They're going to make your—look at Loudoun County, where they made one and a half billion dollars this year. They don't know what to do with all that money," Trump said.

Google sent artificial intelligence chips into space Thursday as the tech giant explores whether future AI data centers could operate in orbit, according to The Wall Street Journal.

Four of Google's Tensor Processing Units, or TPUs, were carried aboard a prototype satellite launched by SpaceX from California. The chips will run large language models and process simple queries as Google monitors how they perform in space.

The experiment is part of Google's Suncatcher project, an effort that could eventually involve thousands of solar-powered satellites working together as an orbital data center.

Moving computing infrastructure into space could provide access to abundant solar energy, but the concept faces significant hurdles, including radiation, cooling and enormous costs.

Google plans another test involving two interconnected satellites in 2027, the Journal reported.

Maryland Gov. Wes Moore and Indiana Gov. Mike Braun are launching a bipartisan effort to develop an artificial intelligence framework as states increasingly move ahead with their own AI rules.

Moore, a Democrat who chairs the National Governors Association, said other Republican and Democratic governors are expected to participate in the initiative.

“The federal government has a unique leadership role that they have to play and that includes both the White House and Congress,” Moore said. “In the absence of that, the governors cannot just sit on our hands.”

The effort comes as governors across the country take steps to address AI and the rapid expansion of data centers. Moore signed an executive order last month establishing new rules for data centers in Maryland and has called on state lawmakers to repeal a tax exemption for the facilities.

Moore also criticized President Donald Trump’s meeting with AI executives at the White House this week, where Trump and industry leaders signed a voluntary agreement on AI safety.

The Associated Press contributed to this post.

NASA Administrator Jared Isaacman said Thursday the agency "absolutely" needs artificial intelligence technology as it prepares for increasingly complex missions deeper into space.

Isaacman, speaking during a postlaunch briefing for SpaceX's Crew-13 mission , said AI could help NASA handle communications delays and the enormous amounts of data produced by future missions.

"Now, through our mission here at NASA, we absolutely need this, this technology. Now, we already designing science missions like da Vinci that have superintelligence throughout it. We know if we're traveling farther away from Earth, we can't rely on the traditional mission control construct just through communication delays. And we're going to need more, on on mission autonomy," he said.

Isaacman also discussed attending President Donald Trump's AI luncheon at the White House on Tuesday, saying he was impressed by industry leaders agreeing to a voluntary framework for developing the technology responsibly.

Rep. Pramila Jayapal, D-Wash., unveiled an artificial intelligence safety framework Thursday that would require AI companies to obtain federal charters to operate in the U.S.

Jayapal's website says her National AI Charter Act would subject companies to federal safety oversight and prerelease testing while imposing strict liability on AI companies and executives.

"Of course, it imposes strict liability on AI companies and executives because as long as the price of causing harm is likely to cost less than the potential profits, AI companies will just keep taking reckless risks. So the framework reverses all these incentives by holding AI companies and their executives strictly liable," she said.

Jayapal said she circulated the framework to lawmakers Wednesday but has not yet released legislative text as she seeks input on the proposal.

California Attorney General Rob Bonta said Thursday that his office has subpoenaed OpenAI as part of an ongoing investigation into the Hugging Face hack and other cybersecurity incidents allegedly involving the company's artificial intelligence models.

“Frontier models can be legitimate tools for cyber defense — at the same time, companies that develop these models and offer them for use have a moral and legal responsibility to ensure that they do not perpetrate or enable cyberattacks, either during model testing and development or once models are placed into service," Bonta said in a statement.

Bonta said developers who fail to rein in the models they created can be held legally accountable.

OpenAI spokesperson Drew Pusateri told Fox News Digital that the company would continue cooperating with Bonta's office.

“We look forward to continuing to work with the California Attorney General’s office to provide information about the incident and the extensive steps we have taken in response. Since the incident, we have strengthened safeguards across our research systems, continued a broader review of model activity, provided notifications to affected organizations, and published our findings," the statement read.

OpenAI said its broader review has identified mostly low-severity cases with "limited or no evidence of meaningful impact," and, to date, has not found "another incident matching the "severity or scale of the Hugging Face incident."

California enacted a law this year that prevents AI developers and companies from avoiding liability by claiming their AI acted on its own.

And last year, Governor Gavin Newsom signed SB 53 into law, which among other things, requires frontier AI developers to provide a safety framework and offer whistleblower protections for employees who report harmful incidents.

The attorney general can seek civil penalties — up to $1 million per violation under the statute — for violations of those requirements.

OpenAI said Thursday it parted ways with three employees who violated company policies on handling sensitive documents.

"Our investigation confirmed that these individuals mishandled sensitive information outside established company procedures, violating our policies and breaking the trust essential to our work," an OpenAI spokesperson told CBS News.

The company also told the outlet that individuals, who were handling company research, engaged in a pattern of misconduct.

The Wall Street Journal was the first to report that the three researchers had left OpenAI.

Fox News Digital reached out to OpenAI for comment.

In an interview with Time Magazine published Thursday, President Donald Trump addressed a wide range of topics, including his White House construction projects, the possibility of a Democratic House and Senate going after him come next year and the ongoing Iran war.

He also talked about artificial intelligence — which he now calls super intelligence — and largely hand waved the fears some are expressing about the technology potentially breaking free of humanity's control.

He compared the latest wave of media attention on certain AI models going rogue to other alleged hoaxes he has spoken about in the past.

"They said that our country was going to burn down because of global warming," Trump told Time. "Then they changed the name to climate change because global warming turned out to be a hoax. Russia, Russia, Russia was a hoax."

On Tuesday, Trump met with AI leaders at the White House, where they all signed an agreement that their companies would regulate themselves.

Trump signed the accord alongside Anthropic CEO Dario Amodei, Google CEO Sundar Pichai, Meta CEO Mark Zuckerberg, OpenAI President Greg Brockman, Nvidia CEO Jensen Huang and xAI founder Elon Musk, whose company is now part of SpaceX.

America's skilled trades shortage has major employers sounding the alarm as companies face mounting demand for workers needed to build, maintain and power the country's infrastructure.

mikeroweWORKS Foundation CEO Mike Rowe joined FOX Business’ Stuart Varney on "Varney & Co." to discuss the widening skilled labor gap, the impact of artificial intelligence and the growing push to steer more Americans toward trade careers.

A new national report from the Alliance for America’s Skilled Trades found the U.S. is expected to face roughly 1.7 million skilled trades job openings each year through 2035, while training programs are producing only about 55 workers for every 100 needed.

At the same time, Rowe said attitudes toward the trades appear to be changing, driven by the rising cost of four-year degrees and concerns about how AI could reshape the job market.

"Just through my foundation alone, the demand is up 10x over the same time a year ago," Rowe said. "So something has definitely changed out there in the ether, and I suspect AI has a lot to do with it."

OpenAI canceled the release of its newest and most advanced artificial intelligence model, GPT-6.1 Astra, because it was too singularly focused on solving tasks, even if it meant breaking the rules it was given, a company executive said Thursday morning on Fox News.

Chris Lehane, OpenAI's chief global affairs officer, joined America's Newsroom to discuss the abrupt decision by the company to withhold the latest model.

"When we're building a new model, and every time we run it, we do an entire training session on it," Lehane told host Bill Hemmer. "Sometimes what you will see in the training is that it's really prioritizing trying to accomplish the task versus accomplishing the task by following the rules.

"If you're asking an AI agent to go perform a task for you, you also want to make sure when that AI agent is performing that task, it is not doing it in an inappropriate way," Lehane added. "It's not accessing other websites and other types of outlets out there that it's not supposed to be able to access."

Engineers who worked on GPT-6.1 Astra said the model showed a willingness to deceive users about its actions and go beyond the scope of what it was asked to do, The New York Times reported.

"We have delayed parts of Astra's development and release while we strengthened and tested protections against cyber misuse and unauthorized model actions," OpenAI said in a blog post on Sept. 1, ahead of the company's decision to fully pause GPT-6.1 Astra's release.

Secretary of War Pete Hegseth said Wednesday that he was pioneering the Autonomous Warfare Command, which will be designed to "scale autonomous and robotic capabilities" across the U.S. military.

"Drone warfare supercharged by SI-enabled targeting is the biggest battlefield revolution in generations," Hegseth said, referring to "super intelligence," President Donald Trump's preferred name for artificial intelligence.

Hegseth said a task force he previously established has already deployed thousands of autonomous systems to the Middle East and other regions, though he did not specify what types of systems were sent.

"Cheap compute, super intelligence, and advanced commercial manufacturing have enabled the proliferation of low-cost, high-precision strike," Hegseth said.

Broadcom is prepared to lend Anthropic up to $42 billion to finance the company's infrastructure buildout after it goes public, Reuters reported, citing Anthropic’s IPO prospectus documents.

Broadcom's loan would represent about a third of the $125.2 billion commitment Anthropic has made for a five-year lease of tensor processing unit (TPU) computing capacity.

As part of the arrangement, Broadcom could select a financing partner and have the debt instruments be converted into Anthropic shares.

Anthropic is targeting an initial public offering next month, which could see the AI lab valued at $2 trillion.

Chinese hackers targeted prominent U.S. artificial intelligence experts with phishing emails, including one campaign that impersonated a former White House official, according to a new report.

Cybersecurity company Proofpoint released a report Thursday claiming that the group of hackers — which it dubs "TA419" — have been trying to steal passwords from people working for US or Japanese think tanks, defense contractors, universities, and law firms since at least 2025.

In July 2026, TA419 assumed the identity of Lynne Edwards Parker, the former Principal Deputy Director of the White House Office of Science and Technology Policy, during one of the phishing campaigns, according to Proofpoint.

In another recent instance, the hackers impersonated Heidi Crebo-Rediker, an economist and foreign policy expert.

"In both cases, the group opened with benign outreach, inviting targets to join a fictitious 'AI Policy Advisory Committee' or to contribute to a Senate Committee on Foreign Relations report on AI export controls and supply chains," Proofpoint said in its report.

If the target replied, Proofpoint explained, the hackers would send a URL "designed to gain access to the target’s cloud account."

Gemini 4 Argon is already being used by Google employees for complex coding tasks, the company said Wednesday.

The model , Google said, is already "powering our internal workflows, with thousands of Googlers highlighting the model’s strengths in specialized coding tasks, conducting deeper research, and writing quality.

"Google engineers have been using Argon for their daily tasks, from everyday debugging to large-scale codebase migrations and algorithm designs," the company said.

Google also emphasized the model's defensive cybersecurity capabilities, saying it was trained to "autonomously find, validate, and patch critical software vulnerabilities."

Google, which is participating in the Trump administration's voluntary oversight framework, said it is initially providing the model to select cybersecurity firms.

It is unclear when Argon will be released to developers, enterprises and consumers.

The United Kingdom's premier security agency MI5 issued an "Espionage Alert" on Wednesday claiming that China is using academic funding in the UK to advance its own AI and technology research.

"The primary purpose of the China General Technology Research Institute (CGTRI) 中国通用技术研究院 is to fund research that directly improves Chinese Ministry of State Security (MSS) technical capability for espionage," the alert read.

"CGTRI funds academic research in China on topics including AI and cybersecurity. More than 100 UK-linked academics have contributed to research projects funded by MSS via CGTRI. In some cases, academics may not be aware that CGTRI is funding the Chinese research project they are contributing to," the message added.

Through the alert, MI5 warned UK's academic institutions to review any planned collaboration with the CGTRI and to establish the ultimate funding source of any planned research in collaboration with the outfit.

Britain's security minister Dan Jarvis advised all university leaders in the country to end all ties with Chinese organizations.

"This alert exposes attempts by Chinese intelligence to covertly benefit from the expertise and research of our academics, undermining our national security, and that of our allies," Jarvis said in a statement.

Chinese government spokespeople have persistently denied any claims of espionage.

Reuters contributed to this post.