Amazon is also alleged to be behind the restrictions imposed on Anthropic's Fable 5 model. According to reports, Amazon CEO Andy Jessi informed senior US administration officials that the Fable 5 model, Claude, provided information that could be used in cybersecurity attacks through certain queries.
According to Semafor, Jessi appealed directly to the government about this without informing Anthropic's management in advance. After that, Anthropic's head Dario Amodei contacted officials and stated that the problem discovered was not a full "jailbreak", but a limited-access vulnerability.
However, David Sacks, a former US administration artificial intelligence advisor, took a different position. According to him, the government demanded that Anthropic fix the problem or withdraw the model, but the company refused.
In response, Anthropic said that the mentioned capabilities are also available in other open AI models and that special restrictions do not need to be overcome for this.
The incident is notable because Amazon has invested about $13 billion in Anthropic and is one of the company's major infrastructure partners. Therefore, some experts see the incident as a clash of strategic interests in the technology sector, as well as a security issue.
