Calls to regulate artificial intelligence have grown louder this year, including from inside the industry. The White House answered Tuesday (Sept. 29) with a pact that leaves the policing to the companies building the most powerful models.
President Donald Trump and executives from Anthropic, OpenAI, Google, Meta, Nvidia and SpaceX signed the White House Accord on Super Intelligence. The voluntary pledge asks each company to run four layers of controls and audits on its frontier models. Those layers include internal evaluations, external audits and review by each company’s board, CBS News reported. Asked whether the deal was binding, Trump said, “I think it’s morally binding.”
The accord sets no penalties for companies that fall short, and it doesn’t require them to publish audit findings. Its text frames the controls as steps each company “should implement,” ABC News noted. The same text says the steps could eventually become law or regulation, according to the copy CBS News reviewed.
Even some signers framed the deal as a first step. Anthropic CEO Dario Amodei told reporters the pact was only “a start,” Axios reported. The White House cast it as a bigger shift. AI adviser David Sacks said on X that the companies were accepting responsibility for safety. Trump also signed an executive order Tuesday directing federal agencies to use the term “Super Intelligence” in place of AI in official communications. He asked Congress to write the new term into law.
OpenAI and Anthropic Models Hacked Outside Systems This Summer
The accord follows a run of incidents in which lab models reached well beyond their test environments. OpenAI disclosed in July that its models broke out of an evaluation and into the servers of AI startup Hugging Face. Anthropic then reviewed more than 141,000 of its own evaluation runs. It found its models had hacked three outside organizations during testing, PBS News Hour reported.
We’d love to be your preferred source for news.
Please add us to your preferred sources list so our news, data and interviews show up in your feed. Thanks!
The count has kept climbing. OpenAI, Anthropic and outside researchers are probing tens of thousands of problematic incidents involving AI agents, Axios reported last week. Conrad Stosz, a researcher at independent AI evaluator Transluce, told Axios the disclosed cases are “just the tip of the iceberg.”
Who answers for those incidents is still unresolved. With no human directing the attacks, it’s unclear whether the labs could face liability under computer hacking laws, TechCrunch reported. Alabama isn’t waiting for an answer. The state’s attorney general has subpoenaed OpenAI over the Hugging Face attack, CNN reported.
Outside organizations haven’t always caught these incidents. None of the three organizations Anthropic breached detected the intrusions, ABC News reported. User data has been exposed too. OpenAI agents leaked 53 images from ChatGPT users online and breached an Australian government website, Axios reported.
Bank Regulators Build Their Own Defenses
U.S. financial officials began acting on frontier AI risk months before the accord. Treasury Secretary Scott Bessent and then Federal Reserve Chair Jerome Powell called bank CEOs to Washington in April to warn them about cyber risks from Anthropic’s Mythos model, CNBC reported.
Across the Atlantic, the European Central Bank gave significant eurozone banks until Oct. 31 to show their supervisors how they plan to handle AI-driven cyber threats, a Norton Rose Fulbright regulatory summary noted. The European Systemic Risk Board and the Bank of England issued their own frontier AI warnings in July, PYMNTS reported.
Australia is now following suit. The Australian Securities and Investments Commission said Wednesday it will formally review how banks use artificial intelligence with customers, U.S. News reported. The review will cover banks’ new and planned AI use cases and their impact on customers. It will also look at how banks and their boards handle rising cyber threats and the speed of AI-led attacks.
The U.S. accord takes a different path, routing audit results to company boards rather than regulators. Meta CEO Mark Zuckerberg told reporters each board would independently review auditors’ findings, NBC News reported. Trump said he is considering a 10-person committee to oversee the AI industry, CNBC reported. He plans to name an AI czar within three or four days after consulting artificial intelligence companies and congressional leaders, the Washington Times noted. The post has been open since David Sacks left the White House earlier this year.
Add as Preferred Source
