U.S. Government Halts Anthropic AI Models Over Security Concerns
- The U.S. government ordered Anthropic to suspend access to its Claude Fable 5 and Claude Mythos 5 AI models, citing national security concerns.
- The directive affects all foreign nationals, including Anthropic’s employees, due to a potential jailbreak vulnerability in the models.
- Anthropic disputes the severity of the vulnerability, arguing it is simple and already replicable using other publicly available models like GPT-5.5.
- David Sacks noted that a trusted partner demonstrated a jailbreak method for Fable, prompting the government’s action after Anthropic refused to fix it.
The U.S. government’s decision to restrict access to Anthropic’s advanced AI models highlights ongoing tensions over AI safety and regulation standards. This move underscores concerns about potential vulnerabilities in cutting-edge technologies that could compromise national security.
By suspending these models, the government aims to mitigate risks while urging Anthropic to address the identified issues promptly before resuming public access. (Source)
