Friday, 12 June, 2026
Claude’s Secret Downgrades Spark Apology
By TechShots Studio
Anthropic has apologized to developers after facing heavy backlash over hidden restrictions in its new Claude Fable 5 model. To address national security risks, the company secretly downgraded or blocked prompts related to cybersecurity, biology, and AI development without notifying users. Admitting they "made the wrong tradeoff," Anthropic has revised its policy. Flagged requests will now visibly fall back to Opus 4.8, and API users will receive explicit refusal reasons.
Read full story at TIMES OF INDIA