Anthropic has halted access to its latest AI models, Fable 5 and Mythos 5, after a directive from the US Commerce Department citing export control restrictions and national security concerns linked to possible jailbreak exploits.
- Fable 5 and Mythos 5 disabled due to export control and security concerns
- Government fears jailbreak exploits targeting sensitive areas
- Anthropic contests recall but complies with legal directives
What happened
Anthropic shut down access to its new large language models, Fable 5 and Mythos 5, shortly after their launch, following a directive from the US Commerce Department. This directive imposed export controls restricting these models' usage outside the United States and required immediate compliance.
The company announced the suspension is necessary to adhere to government orders after reports emerged of a potential jailbreak exploit that bypasses content safeguards, specifically concerning cybersecurity, chemistry, and biology prompts. Access to Anthropic’s other AI models remains unaffected.
Why it matters
The government’s concerns focus on maintaining national security by preventing malicious use of advanced AI technologies. The reported jailbreak could allow users to circumvent restrictions designed to prevent the AI from generating harmful or sensitive content, raising risks in critical areas like cybersecurity vulnerabilities.
Anthropic expressed disagreement with the government’s decision to recall the models based on these narrow jailbreak instances, noting that similar capabilities exist in other commercially deployed models. The move highlights ongoing tensions between AI innovation and regulatory efforts to manage emerging risks.
What to watch next
Anthropic indicated it expects further clarification and updates on the situation within the coming weeks, as government agencies work to enhance security systems against potential AI exploits. This timeline may influence when Fable 5 and Mythos 5 could return or be revised for relaunch.
More broadly, this incident underscores the increased scrutiny AI developers face amid evolving US government policies aiming to balance technological progress with national security. Industry stakeholders will be closely monitoring regulatory responses and potential impacts on AI model deployment strategies.