Inviting hard questions

Anthropic has launched a feedback program that pays external experts to identify potential safety vulnerabilities and adversarial prompts within its AI models. By formalizing this method of red teaming, the company aims to reduce the risk of bias, hallucinations, and misuse before releasing new updates. This shift toward community-driven scrutiny reflects a broader industry attempt to establish more rigorous security benchmarks for increasingly capable generative systems.