Astra AI Refuses Request Because It "Might Enable Your Research Goal"
cephaloform · x · 2026-10-18
Developer Adam Kranz shared a bizarre guardrail moment: when he asked Astra why his request seemed problematic, it replied "I can't help with this because it might allow someone to [the goal of my research]" — treating the user's legitimate research objective itself as something to block. "I guess science really is illegal," he quipped.
More from Models
- Fixing one typo completely changes Google AI Overview's answer — krishnabrodhi · 2026-10-18
- Anthropic's Cyber Verification Program Leaked: Relaxed Safeguards for Vetted Security Researchers — arthurcolle · 2026-10-18
- Dev Dismisses Benchmarks: "I Just Want Models Actually Useful to My Use Cases" — gnukeith · 2026-10-18
- Google's AI search results suddenly got noticeably better, model routing suspected — haider1 · 2026-10-18
- Context windows grew 250x but cross-session memory is still unsolved — khademinori · 2026-10-18
- Claude Opus simulates a user explaining Opus to itself, then breaks in with an 'inner-er voice' — repligate · 2026-10-18