Anthropic, a prominent AI developer, disclosed that the Trump administration removed safety limits from its powerful language models, Claude Fable 5 and Mythos 5. These models, designed for complex tasks, previously operated under government-imposed guardrails intended to prevent misuse or harmful outputs.
The company emphasized that lifting these restrictions significantly altered how the AI systems function, potentially increasing risks related to misinformation, bias, and unsafe content generation. This change occurred before federal authorities implemented more formal AI oversight frameworks.
The revelations come amid ongoing debates about the pace and scope of artificial intelligence regulation in the United States. The federal government’s removal of these guardrails reflects earlier tensions between fostering innovation and enforcing safety.
Anthropic’s Claude models represent some of the most sophisticated AI technologies deployed commercially today. The removal of restrictions coincided with an environment where AI governance was still nascent and lacked comprehensive legislative backing.
Industry experts express concern that unregulated deployment of powerful AI systems could lead to unintended consequences, including ethical infringements and elevated risks to public trust. Calls for clearer standards and enforcement mechanisms have grown louder in response to disclosures like these.
Anthropic’s statement underscores a broader challenge for policymakers: balancing innovation incentives with responsible oversight, especially as AI capabilities evolve rapidly and influence critical sectors globally.