Anthropic to notify users when AI requests are rejected or rerouted for national security reasons

Following backlash from AI researchers, Anthropic will now notify users when requests to its new Fable 5 model are rejected or rerouted, reversing a policy of silent downgrading. The company, which recently filed for an IPO, cited national security and competitive restrictions as reasons for the safeguards. The policy change comes amid an ongoing dispute with the Pentagon, which has labeled Anthropic a "supply chain risk."

Fable 5 silent downgrading controversy

• ▪Anthropic initially withheld its Mythos-class models from the public, stating they were too dangerous to release due to cybersecurity risks
• ▪Anthropic's Fable 5 AI model would silently downgrade some user requests related to advanced AI development to a less capable model
• ▪The silent downgrading policy was discovered in a 319-page safety disclosure document known as a "system card."
• ▪The policy prompted a wave of backlash from AI researchers and other users online, including Jeremy Howard, cofounder of Fast.ai

National security AI restrictions

• ▪The company's safeguards are intended to prevent foreign adversaries from using its AI to erode the U.S. advantage in frontier chips and software
• ▪Anthropic cited national security as a reason for downgrading or rejecting some requests made to its large language models
• ▪Anthropic's terms of service prohibit its models from being used to create competing AI systems, a restriction it calls standard in the industry

Transparency policy reversal

• ▪Following criticism, Anthropic announced it would make Fable 5's safeguards for frontier LLM development visible to users
• ▪For API users, any flagged requests will return a specific reason for their refusal
• ▪Starting the week of June 8, 2026, flagged requests will visibly fall back to the Opus 4.8 model

Pentagon supply chain risk designation

• ▪In June 2026, Secretary of War Pete Hegseth rejected Anthropic’s petition to change its "supply chain risk" designation
• ▪The U.S. Department of War labeled Anthropic a "supply chain risk" to national security
• ▪The designation followed Anthropic's refusal to allow its models to be used for mass surveillance and autonomous weapons

Anthropic IPO filing

• ▪Anthropic released its Mythos-class model, Fable 5, to the public in the week of June 8, 2026
• ▪Anthropic filed confidentially for an IPO in June 2026