AI guardrails impede offensive cybersecurity researchers' work; researchers say OpenAI’s and Anthropic’s guardrails restrict vulnerability discovery and weaponization tools

Original headline: "How AI guardrails are impeding the work of offensive cybersecurity researchers"