Trump Softens Stance on Anthropic, Says AI Firm No Longer Poses Security Risk
Last week, U.S. President Donald Trump hinted that he may have considered the AI firm Anthropic a danger to national security. But now, as revealed in an interview that was released on June 19th by "The Axios Show", he no longer does.
Earlier this week, senior technical personnel from Anthropic were supposed to meet with authorities from the Trump administration. The meeting was planned to talk about a disagreement regarding foreign access to its two most powerful artificial intelligence models, Mythos 5 and Fable 5. Following Trump's directive to Anthropic to restrict access to those models for foreign nationals, the company last week stopped access for all users.
Anthropic Faced Backlash as U.S. Limits Access
In an announcement made on the evening of June 12th, the San Francisco-based business Anthropic stated that Mythos 5 and Fable 5 had been ordered suspended by the US Department of Commerce for "national security" grounds; however, no other information was provided. In contrast to Mythos 5, which had few restrictions and was accessible to all partners, Fable 5 was carefully guarded to avoid any serious abuse.
This prevention was specifically designed to prevent the development of chemical and biological weapons or cybercrime. However, Anthropic claimed that an unnamed organization informed the Trump administration that it had discovered a method to circumvent security measures meant to prevent Fable 5 from being exploited in a cyberattack.
Several news sources have named Amazon as the third party that found the "narrow" vulnerability, and Anthropic has characterised the software vulnerabilities it uncovered as "minor". Anthropic claimed it had to take its models offline because it couldn't differentiate between users based on country, even though the order solely applied to access by foreign citizens. It is rare for a government to blatantly prohibit a local company's advanced AI model. China restricts access to key domestic AI businesses and prohibits access to the most capable Western AI models; nonetheless, these restrictions are typically included in the models before they are released. According to entrepreneur Martin Varsavsky, the order has huge ramifications.
Anthropic Quickly Responded to US Administration’s Order
Trump informed that Amodei handled the export control directive "very quickly" and "responsibly". At a summit in France this week, the US president and other G7 leaders met with tech chiefs, including Amodei. Axios reports that Trump has not ruled out the possibility of invoking emergency powers under the Defence Production Act to take action against Anthropic. For the DPA, "I have the power to use a lot of things," Trump continued.
"But I'm not sure I have to do that." Anthropic expressed gratitude to the administration for their continuous cooperation in resolving this issue expeditiously in response to a question about Trump's interview. The spokesman stated that the AI firm is still dedicated to collaborating with the US administration in order to achieve their common objectives of safeguarding vital infrastructure and maintaining the US's position as the artificial intelligence industry leader.
