The US government recently directed Anthropic to suspend access to the two models over national security concerns, forcing the company to shut them down for users worldwide. Anthropic has said it disagrees with the decision and believes the concerns are based on a limited "jailbreak" technique. According to a Wall Street Journal (WSJ) report, the technique in question was tested by researchers at
Amazon, who used a series of prompts to get the AI model to identify a small number of software vulnerabilities.
Amazon researchers reportedly tested model safeguards #
According to reports, the jailbreak research was carried out by researchers at Amazon. Katie Moussouris, chief executive of cybersecurity firm Luta Security, said the researchers used a series of prompts to make Anthropic's model provide information about a handful of security vulnerabilities.
Anthropic had reportedly shared a copy of the report with her, she said. The company has maintained that the vulnerabilities identified were already known and relatively minor.
"We reviewed a demonstration of this specific technique being used to identify a small number of previously known, minor vulnerabilities," Anthropic said in a statement.
"These vulnerabilities all appear relatively simple, and we have found that other publicly available models are able to discover them as well without requiring a bypass," the company added.
US government restricts Mythos 5 and Fable 5 access
The US government on Friday, June 12 ordered Anthropic to suspend access to Fable 5 and Mythos 5, citing national security concerns. The directive applies to foreign governments, companies and individuals, including foreign nationals inside the United States.
Anthropic said the broad scope of the order forced it to disable both models for all users to ensure compliance. The company said it was not given detailed information about the government's concerns.
"The letter did not provide specific details of its national-security concern," Anthropic said.
Mythos 5 is Anthropic's most advanced AI model and has been used by governments and companies to identify and fix software vulnerabilities. Fable 5, released earlier this week, was designed as a public version of the technology with additional safeguards.