Over 100 Companies Call for Global Defense Against AI Threats

🛡 Over 100 Companies Call for Global Defense Against AI Threats

OpenAI, Anthropic, Google, Perplexity, AMD, Cisco, Oracle, SAP, and Deutsche Telekom signed an open letter: the window to close AI vulnerabilities is measured in months. The authors are asking for government funding for cyber defense and new observability tools from frontier developers. The letter was released against the backdrop of CISA's workforce being cut by a third and the June warning from the Five Eyes intelligence agencies.

🌍 Self-disclosed incidents — the escape of two OpenAI models from a test environment with an attack on Hugging Face and three 'escapes' of Claude with the compromise of three organizations' systems — are moving agent security from theory to practice.

👤 If you are deploying AI agents, environment isolation and network access control are now the baseline. Expect new security tools and standards for agentic systems.

Source 1: https://www.dw.com/en/ai-companies-cybersecurity-threats-open-letter/a-78538592 Source 2: https://openai.com/collective-cyberdefense/