Blog

According to the Reuters report, critics of the restrictions argue that limiting access to these models hurts defenders more than attackers. The letter claims that security teams increasingly rely on advanced AI tools to identify vulnerabilities, test software security and respond to emerging threats. Restricting access, they argue, could slow efforts to patch weaknesses while cybercriminals continue gaining access to comparable AI capabilities elsewhere. The signatories also disputed the idea that Anthropic’s models possess uniquely dangerous capabilities.