Story perspectives

OpenAI, 100+ Tech Firms Urge Governments to Bolster AI Cyber Defense

8/30/2026

Summary

Tech Leaders Push Action

OpenAI and over 100 tech companies issued a letter urging action against AI-enabled cyber attacks.
The letter warns AI-enabled attacks will become widespread and sophisticated within months as models improve.
It flags hospitals, water-treatment plants and internet infrastructure as targets.
Signatories include OpenAI, Microsoft, Google, Anthropic, AWS, Cisco, CrowdStrike, Palo Alto Networks.
The letter urges governments to fund cyber defense, expand programs and give AI tools to hospitals and utilities.

Summary

Five Eyes Warn

The Five Eyes council warned that frontier AI models could breach global cyber defenses within months, on June 22.
Keven Knight, CEO of Talion Cyber Security, said China is developing models comparable to Mythos and could target the West.
Andrew Jones, co-founder of Adaptive Security, called the Hugging Face breach “clear evidence” of a fully autonomous AI cyberattack.

History

Historical Context

First identified on 17 June 2010, Stuxnet was a Windows worm that uniquely attacked SCADA systems at Iranian nuclear facilities, with development traced to 2005-2006. The same codebase produced the Duqu worm later in 2010, showing how advanced malware can evolve and prompting warnings that modern AI models could enable comparable attacks on hospitals and utilities.

1 of 3

Tech Leaders Push Action

OpenAI and over 100 tech companies issued a letter urging action against AI-enabled cyber attacks.
The letter warns AI-enabled attacks will become widespread and sophisticated within months as models improve.
It flags hospitals, water-treatment plants and internet infrastructure as targets.
Signatories include OpenAI, Microsoft, Google, Anthropic, AWS, Cisco, CrowdStrike, Palo Alto Networks.
The letter urges governments to fund cyber defense, expand programs and give AI tools to hospitals and utilities.

1 / 3