OpenAI, Google and 116 entities sign letter warning of limited window to address AI-powered attacks against critical infrastructure.
Published
OpenAI, Google and dozens of other tech companies have signed a letter calling for urgent government action against AI-powered cyber threats, with 116 companies and entities now backing the initiative. The letter warns of a "limited window" to deploy coordinated defenses and urges making cyber defense accessible to critical infrastructure under pressure.
The joint push represents one of the largest industry-wide calls for AI security coordination to date. The letter specifically urges a coordinated government effort to help protect critical systems from increasingly sophisticated AI-driven attacks.
Separately, Anthropic announced a new Model Hardware Standard designed to help AI agents operate physical machines and machinery. The standard is initially available as a research preview, with plans to open source it in the future.
OpenAI is also developing a "persistent" AI agent called Codex, according to code reviewed by WIRED. The feature would enable the AI to continue working proactively until explicitly "put to sleep" — representing a shift toward more autonomous AI systems that can operate with less human intervention.
Sources: 1. CNBC: 'We have a limited window': 116 companies, entities sign on to major AI cyber defense push — https://www.cnbc.com/2026/08/27/ai-cyber-defense-letter.html (src1) 2.