Anthropic boss Dario Amodei urged world leaders to “resist the temptation to splinter” in their approaches to AI regulation – even as his employees reportedly posted internal work chats griping that the White House was unfairly targeting them.
The plea came during a Wednesday lunch at the G7 Summit attended by President Trump, OpenAI CEO Sam Altman and other tech leaders who gathered in Évian-les-Bains, France – just days after the Trump administration slapped export controls on Anthropic’s “Fable” AI model due to cybersecurity concerns.
Amodei told attendees that democratic countries should work together on AI, including efforts to prevent bad actors from gaining access to advanced models, the Financial Times reported, citing sources close to the discussions.
Altman — a longtime nemesis of Amodei, who left OpenAI in December 2020 to found Anthropic — nevertheless backed Amodei’s point, saying G7 countries should have access to AI-powered cybersecurity tools.
Demis Hassabis of Google DeepMind also voiced support for more collaboration, according to the report.
Leaked internal chats reveal that Anthropic employees were taken aback by the shutdown – with some reportedly asking if it would hurt the company’s plans to go public, according to leaked chats obtained by the New York Times.
In one message, an employee questioned if Anthropic was being “bullied” by the Trump administration “based on bad vibes,” according to the report.
“At what point does this just feel like they don’t want us to exist?” another employee asked in an internal chat on Tuesday.
French President Emmanuel Macron told G7 attendees that the Trump administration’s action against Anthropic had “clarified the stakes” and warned that leading AI developers could suffer if the government “from one day to the next can turn off the switch,” according to the FT.
An Anthropic spokesperson declined to comment on Amodei’s reported remarks.
The Trump administration cracked down on Anthropic late last week after it received intel from Amazon CEO Andy Jassy and others suggesting its advanced AI models had flaws that could hurt national security. Anthropic has argued those concerns are overblown, but shut down the models entirely in order to comply with export controls.
The Trump administration’s abrupt action against Anthropic sparked concern about other G7 nations who suddenly found themselves without access to Mythos – which the company has said has unprecedented capabilities to find and exploit software vulnerabilities.
On Tuesday, The Post reported that the Trump administration wouldn’t allow G7 nations to regain access to Anthropic’s models, with one official describing the notion of granting carve-outs to export controls as “completely illogical.”
Members of Anthropic’s technical staff have been meeting with Trump administration officials for days in a bid to resolve their safety concerns.
Anthropic has said it is committed to getting the situation resolved as quickly as possible.
Senior administration officials previously told The Post that they were miffed by Anthropic’s effort to downplay the flaw as a “narrow potential jailbreak” – after Amodei and others have spent the last several years raising alarms about the dangers of AI.
“They seem obsessed with safety for everyone except themselves,” one senior Trump administration official told The Post. “If I was an Anthropic investor, I would be extremely concerned.”
