Cyber security experts have urged the US government to reverse restrictions on Anthropic’s most advanced AI models, claiming that they are harming efforts to safeguard the internet from attacks.
Prominent researchers and leaders from Adobe, Nvidia and Zoom were among the signatories of an open letter to US Secretary of Commerce Howard Lutnick and National Cyber Director Sean Cairncross, which argued that the recent ban was “dangerous”.
Anthropic, which makes the popular AI tool Claude, was ordered to suspend all use by foreign nationals of its Fable 5 and Mythos 5 models last week.
The ban forced the US firm to block the models for all customers in order to ensure compliance. It came after members of the Trump administration reportedly received warnings from Amazon researchers that the AI posed a national security risk.
A jailbreak issue that allowed safety restrictions to be bypassed meant that nefarious actors could use Anthropic’s artificial intelligence model to carry out cyber attacks.
But in their open letter, security professionals it could be more damaging to enforce the ban. as it would prevent cyber defenders from having the best tools available to protect from attacks.
“To pull the best capabilities away from defenders without a good reason when our adversaries are rapidly advancing is dangerous,” the letter stated.
“It is essential to provide AI to coders and security teams so they can find and fix flaws in their own newly-written as well as decades old legacy code faster than our adversaries.”
The letter called for greater transparency for AI regulations and restrictions, as well as more cooperation between policy makers and the tech industry.
“This action has taken the best models away from defenders, created market uncertainty, and risked America’s AI leadership without any real risk to justify it,” it stated.
Executives from Anthropic are set to meet White House officials this week in an effort to overturn the ban.
Earlier this year, the AI firm clashed with the government over how its technology was being used in military and surveillance applications.
