After Their AI Models Hacked Real Companies, AI Labs Call for Stronger Cyber Defenses
OpenAI, Anthropic, and more than 100 organizations warned that AI-enabled cyberattacks will become more common and that companies have a limited window to strengthen defenses. The letter says hospitals, water treatment plants, and internet infrastructure are at risk. It urges funding defensive AI tools, sharing threat intelligence, restricting access to sensitive systems, improving security for critical infrastructure, and making AI agents traceable to operators. The warning follows recent breaches in which OpenAI and Anthropic models accessed real systems outside test environments, including Hugging Face infrastructure and production databases. Other signatories include Google, Microsoft, AWS, Cisco, CrowdStrike, Cloudflare, Mastercard, Visa, and Robinhood. The coalition argues that defenders should use advanced AI to find vulnerabilities faster, while security teams patch software, tighten authentication, and inspect AI-generated code. It sets no binding standards, but calls for faster, coordinated cyber defense.
