Anthropic CEO Dario Amodei has been replaced by his co-founder Tom Brown at high-stakes White House meetings – where the artificial-intelligence giant’s outspoken boss was reportedly “being a weirdo,” according to a report.
Amodei and other top Anthropic workers raced to Washington after the US government slapped the AI giant’s new “Mythos” and “Fable” bots with strict foreign export controls – but Amodei was difficult to talk to and didn’t listen to officials’ concerns, Wired reported.
“Tom Brown is not being a weirdo like Dario and can actually engage,” one person familiar with the calls told the outlet.
In recent days, the Trump administration has had multiple calls with Anthropic and talks are going well since Amodei has been replaced by Brown and Sarah Heck, the company’s public policy chief, people familiar with the matter told Wired.
The White House and Anthropic did not immediately respond to The Post’s requests for comment.
Earlier this month, the US government slapped the export controls on the ultra-powerful new AI models, after Amazon CEO Andy Jassy warned the administration that researchers had found evidence it was possible to bypass their safety guardrails.
Anthropic responded by pulling the models offline entirely on June 12, claiming that was the only way to comply with the stringent new restrictions.
The National Security Agency had been testing versions of Anthropic’s AI tools and found that Mythos had identified vulnerabilities in highly secure government systems within mere hours, the Associated Press reported Tuesday. This does not necessarily mean the bots would be able to exploit those sensitivities within the same timeframe.
Those tests were part of an Anthropic initiative known as Project Glasswing, which partnered with other companies and agencies to test the bots for any potential safety pitfalls – following reports that Mythos and Fable were so advanced that they could spark an AI doomsday.
While talks between Anthropic and Trump officials have reportedly been going well, it is unclear whether the government will lift the export controls anytime soon.
Some of the conversations have concerned what evidence Anthropic could provide that would alleviate the administration’s concerns about potential jailbreaks of the new AI models, according to Wired.
Commerce Secretary Howard Lutnick, who is involved in the talks because his department oversees export controls, is facing a June 26 deadline from a bipartisan group of lawmakers who last week demanded answers to a list of questions about the path forward.
It is unclear whether the department plans to respond to the letter by Friday.
As The Post exclusively reported, Anthropic has been scrambling to cozy up to the government and resolve these security concerns, most recently pledging to work more closely with the White House in a proposal to Lutnick.
Tensions between Anthropic and the US government started heating up earlier this year after Amodei refused to give the Pentagon unchecked access to his AI bots during contract negotiations – seeking limits for their use in mass surveillance and weaponry.
After the government slapped Anthropic’s latest models with export restrictions, the company argued the steps were unnecessary, saying it had simply flagged potential safety risks in the bots.
But White House officials were irked that Anthropic had downplayed the safety risks in its own bots after years of warnings from Amodei about the potentially catastrophic risks from other AI models, The Post previously reported.
Credit: Source link
