AI Safety Filters Are Slowing Down Legitimate Cybersecurity Research

Offensive cybersecurity researchers, who identify unknown vulnerabilities and build exploit tools, say AI guardrails from major providers are interfering with their work. TechCrunch interviewed multiple professionals in this field about the specific impact of restrictions imposed by OpenAI and Anthropic. These safety filters, designed to prevent misuse of AI, are also blocking legitimate security research tasks. Researchers suggest the broad nature of these guardrails fails to distinguish between malicious actors and professional security experts.