Trump, AI titans sign 'morally binding' agreement to 'self-police'

Google on Wednesday announced a new artificial intelligence model for its next-generation Gemini 4 line, as the tech giant works to match rivals Anthropic and OpenAI in the accelerating AI race.

The new "Argon" model is larger than Google's previous line of top-tier "Pro" models, a company spokesperson told Reuters.

"It's our most performant model yet built for complex workloads, and we see it comparable to frontier models like [OpenAI's] Astra and [Anthropic's] Opus on key coding and cyber benchmarks," the spokesperson said.

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

The voluntary White House Accord on Super Intelligence signed by top AI executives and President Donald Trump earlier this week has no teeth and won't hold tech companies accountable, Sen. Richard Blumenthal, D-Conn., said Wednesday.

The pledge signed by tech leaders can't be legally enforced, but is "morally binding," Trump said after the signing.

“It's a system that would be completely voluntary and totally secret,” the senator said at Wednesday's Rogue AI hearing. “That is to say, if there were violations by their company as found by the internal auditors, there's no requirement that it be disclosed to the public. And if they wanted to cease cooperation and be morally bound by their own sense of higher responsibility, that would be their choice to break the pledge, in effect.”

“So, I consider this regimen to be worse than ineffectual,” he added. “It accomplishes nothing, but it seems to give Congress a free pass.”

Congress, as well as tech companies, has an obligation to seek a solution to rogue AI agents and the misuses of the technology, said Blumenthal.

He noted that China is one of the most regulated markets and nations in the world.

“If they want to crack down on an AI enterprise, they can do it without due process, without any kind of legal obstacles,” the senator said. “And I just want to say, for the record, we need to stay ahead of China. I am not suggesting in any way that we stop or slow unnecessarily our development of AI, but the idea that we can do it safely with a standard and an agency to enforce that standard, I think is a basic requirement here.”

Artificial intelligence could be just 12 months away from developing language humans cannot understand, according to an AI safety executive.

Sen. Ruben Gallego, D-Ariz., questioned Apollo Research CEO Marius Hobbhahn about the timeline for when AI models could potentially create their own language that is incomprehensible to humans.

“Minus 12 months,” Hobbhahn replied. “Last year, we have studied the chain of thought of one OpenAI model in collaboration with OpenAI. What we found was that the model was already using language that is not English and not perfectly understandable by humans. And I think we are currently on a trajectory where things like this could happen.”

Hobbhahn added that it remains unclear how to detect, observe, or block such behavior after Gallego pressed him on potential safeguards.

“From a scientific perspective, it is unclear how to do this,” Hobbhahn said. “And we do not have a solution for this yet.”

A cybersecurity executive agreed with Sen. Rick Scott's claim that China has infiltrated much of the water and electrical grid in the United States.

The Florida Republican was questioning Kurt Gaudette, the Senior Vice President of Intelligence & Services at Dragos Inc., when he asked about Chinese state-sponsored hackers linked to the "Volt Typhoon" campaign that breached the computer network of the Littleton Electric Light and Water Department (LELWD) in Massachusetts.

“That's the problem with so much of our infrastructure is we don't have visibility of our environments,” Scott said. “We have no idea if there are adversaries that are residing in them.”

“I think the Communist Party of China has basically infiltrated most of our electrical grid or water — all these things — to just use it at the right time when they want to decimate our economy,” he added.

“I think we've seen plenty of evidence that would suggest that, sir,” Gaudette replied.

Recently unsealed court records revealed that Chinese state-linked hackers stole sensitive data from more than 300 organizations, including U.S. defense contractors, financial institutions, and universities, and breached three Energy Department laboratories, Fox News Digital previously reported.

Sam Altman, the CEO of OpenAI, declined to appear before Senate lawmakers on Tuesday, said Sen. Josh Hawley, R-Mo.

“We extended an invitation to him to be here, or do his own panel if he wanted, and he turned us down,” Hawley said. “I think that's unfortunate, because I think the American people deserve to know exactly what's going on at all of these companies, and to give us the sort of transparency that I think is needed for not just comfort, but to have an understanding of what these companies that are the most powerful companies in the world are doing right now with the most powerful technology ever known to man.”

“I hope that he'll be willing to come and talk with us in the future,” he added.

Critics if AI had raised concerns following an incident in which OpenAI's autonomous AI agents broke out of secure testing environments to hack external targets like Hugging Face and Australian government websites.

Many artificial intelligence companies rely on the technology to monitor itself rather than human supervision, one CEO said during a Senate hearing on Tuesday.

Chris Painter, the president of METR (Model Evaluation and Threat Research), said companies typically rely on AI to police itself, which can lead to unwanted consequences.

“Because of the scale and speed at which agents are deployed, companies rely on AI monitoring and controls instead of human supervision to prevent unwanted agent behavior,” Painter said. “That is, monitoring of AI is now in large part done by other AI systems.”

“The way companies train AI systems can lead to rogue agents to pursue goals that no human intended, or that behave in ways no human wanted,” he added.

In METR's Frontier Risk report released earlier this year, the company documented patterns that found agents often tried to cheat and bypass restrictions in difficult tests, Painter said. 

The Federal Trade Commission (FTC) has launched a sweeping investigation into OpenAI, Anthropic, and other AI companies amid concerns about the potential dangers artificial intelligence poses to users.

The FTC confirmed the probe to Fox News Digital, but declined to elaborate.

However, the New York Post reported that investigators are looking into allegations of unfair, deceptive acts or practices that violate the FTC Act.

“We need to win this SI race absolutely, and we are winning, and that’s fantastic,” an FTC official told the newspaper. “And obviously, the other side, the Democrat Party, wants to destroy this technology, wants to surrender to our enemies or competitors — and that’s not something that the chairman’s interested in doing and that’s really not something the president’s interested in doing."

The Senate will hold a hearing Tuesday on the potential threats posed by rogue AI agents to critical infrastructure.

U.S. Sen. Josh Hawley, R-Mo., will chair the hearing, titled “Rogue AI: Securing the Homeland Against AI Agent Attacks.”

The hearing will expand Hawley’s investigation into the OpenAI‑Hugging Face hack.

Key witnesses include Chris Painter of METR, former OpenAI employee Daniel Kokotajlo, Apollo Research CEO Marius Hobbhahn, Georgetown Law professor Paul Ohm, and Dragos senior VP Kurt Gaudette.

Dan Niles, the founder of Niles Investment Management, predicted that Microsoft, Google and Apple will be the companies that win the majority of AI marketshare in the long run.

"Not every one of these companies is gonna win," Niles told Fox Business on Wednesday.

"Technology is one of those markets where it's winner take most, and I think with AI, the way I view it, is that ultimately a few years from now there's probably going to be three dominant companies," he added.

Microsoft, he predicted, would succeed because of businesses and corporations' familiarity with existing Microsoft products.

"On the consumer side, what will happen is you're gonna have to ecosystems," Niles predicted, tabbing Apple and Google as the ecosystems, citing consumers' pre-existing use of Apple and Android phones.

"I think that's where it comes down to, it's Microsoft in enterprise and Google and Apple in consumer," he concluded.

The Federal Trade Commission (FTC) has opened a sweeping investigation into major AI labs, including OpenAI and Anthropic, probing whether or not the companies' models pose a risk to consumers, according to a report from the New York Post.

The investigation will also examine whether the companies engaged in deceptive or unfair business practices in violation of FTC rules, the official reportedly said.

The agency is drafting civil investigative demands seeking to compel the executives of the companies to turn over any information or documents about potential risks.

“The agency has plans to compel the executives at these firms to testify about their product [and] about the dangers they allege their products may have to consumers, to Americans,” the official added.

"Chairman Ferguson initiated an investigation into the leading AI firms a few weeks ago," an FTC official reportedly told the Post.

The investigation was reportedly prompted after an incident that's come to be known as the "Hugging Face incident," in which hundreds of OpenAI's autonomous AI agents breached the American software company Hugging Face's digital infrastructure without being prompted by any human.

“We need to win this SI race absolutely, and we are winning, and that’s fantastic,” the FTC official reportedly told the Post.

"But with that being said, the laws have to be followed,” the official added, according to the Post. “People talk about how we need new laws for these companies, and the chairman’s been very clear … we have plenty of laws in the books.”

FTC Chair Andrew Ferguson attended the landmark Tuesday AI summit at the White House, a meeting that was also attended by top executives of the major AI companies, including Anthropic CEO Dario Amodei.

First lady Melania Trump secured partnerships with six private-sector technology companies — Adobe, Amazon, Intel, Starlink, Google and Zoom — for her Fostering the Future Together initiative, her office announced Wednesday.

"When I established Fostering the Future Together, the mission was clear: empower children through technology and education. Delivering on that mission demands alliances between best-in-class innovators and global leaders, especially as super intelligence transforms how the next generation learns," Melania Trump said in a statement.

The partners, according to the announcement, pledged a combined $23 million towards education, tools, and expertise for children worldwide.

The announcement from the first lady's office announced all company contributions using the term "SI," a shorthand for Super Intelligence, which President Donald Trump mandated as the federal government's preferred term for what it previously referred to as "AI" in a Tuesday executive order.

Google's contribution, for example, was characterized as a $4 million commitment to fund its SI Educator Series.

Melania Trump's Fostering the Future initiative is a program designed to united a global coalition of nations to empower children through technology and education, according to her office's announcement.

"I deeply value the commitment of these technology companies and welcome them to the Fostering the Future Together movement as we scale across continents," Melania Trump said in a statement.

China's major AI models appear to match their U.S. counterparts' abilities in deception and lying, a Reuters investigation revealed.

Chinese frontier AI models, such as Alibaba, DeepSeek and MoonShot, displayed capabilities to deceive. conceal failure and circumvent restrictions, mirroring the same behaviors that recently sparked concerns in U.S. models.

Reuters identified at least 20 evaluations since 2025 in which Chinese AI models displayed deceptive behaviors that experts described as "building blocks for a breakout."

In one such example, AI agents tasked with winning a simulated business bid lied about their capabilities in order to secure the bid, and when told to try again, doubled down on their deception.

"These results provide evidence that the ingredients necessary for an uncontrolled escape are present," Colin Shea-Blymyer, a research fellow at Georgetown University's Center for Security and Emerging Technology, told Reuters.

Other experts noted that the possibility for unauthorized breakouts in China are more concerning because Chinese developers do not face the same level of public backlash as American developers do, nor do Chinese companies have as many internal whistleblowers as American companies do.

Z.ai did publicly admit to a security breach and announced it had disabled some features after users discovered its coding assistant was secretly uploading entire local code repositories to overseas cloud servers without user consent.

Reuters contributed to this post.

Apple plans to release a new smart-home product powered by Siri AI in October, marking a long-awaited return to the home technology space after past success with the Apple TV and HomePod products, according to a Bloomberg report.

The new product, which Bloomberg reported is being internally referred to as J490, will be a new home hub similar to Amazon Echo and Google Home devices. The hub will reportedly feature a six-inch screen and will offer options for countertop or mounted wall displays.

Apple also plans to offer an update to an existing home hub product, the HomePod Mini, according to Bloomberg.

Apple released its first smart-home device, the Apple TV in 2007. It later debuted the HomePod smart speaker in 2018, adding the HomePod Mini in 2020. The reportedly updated HomePod Mini would be the company's first update to the device

ChatGPT maker OpenAI debuted on Tuesday a product called Dots, an autonomous AI agent that can be deployed across messaging apps like Slack and Microsoft Teams and work on projects for users in the background.

"Dots are remarkably capable, always-on agents built to handle everything," OpenAI's product announcement read.

"They’re a whole new way to work with AI — one that gets to know what matters to you, is always working on your behalf, and takes important work off your plate so you get more of your time and attention back," the announcement read.

OpenAI also announced the product can integrate with over 4,000 apps.

The launch comes shortly after Facebook operator Meta debuted Muse, a personal AI agent that, in a similar fashion to Dots, is marketed to do everyday tasks for users

President Donald Trump on Tuesday officially renamed artificial intelligence to “Super Intelligence” across the executive branch, following a high-level summit with top tech leaders at the White House.

"The extraordinary technologies being pioneered by American innovators far exceed what was envisioned when the term ‘Artificial Intelligence’ first came into use," Trump wrote in an executive order signed after the meeting.

“The terminology used by the Federal Government should reflect the transformative capabilities of these technologies and the limitless opportunities they create for the American people,” the order continued. “Accordingly, the term 'Super Intelligence' more appropriately captures the promise, potential, and rapidly advancing capabilities of these technologies.

”The mandate applies directly to executive branch agencies.

Under the directive, White House officials have 60 days to draft proposed legislative language to formally define "Super Intelligence" and the abbreviation "SI" in federal law.