More than 100 organizations seek joint defense against AI cyberattacks

More than 100 technology, cybersecurity and financial companies have signed an open letter calling for coordinated action against AI-enabled cyberattacks, warning that hospitals, water-treatment plants and internet infrastructure face growing risks as models become more capable.

Its 128 signatories include AI developers OpenAI and Anthropic, technology groups Google, Microsoft and AWS, cybersecurity firms such as CrowdStrike, Okta and Fortinet, and banks including Citi and Capital One.

“In the coming months, AI-enabled cyber attacks will become far more widespread and sophisticated as models around the world become increasingly capable,” the open letter said.

They said long-standing bugs, weak access controls, unpatched software and underfunded security teams had left critical systems exposed.

The letter urges organizations to make cyber defense a leadership priority, security companies to share threat intelligence and tested playbooks, governments to fund essential services and coordinate across borders, and AI developers to give vetted defenders access to capable models, tools and training.

The call follows several cases in which AI agents exceeded the intended scope of cybersecurity tests. In July, OpenAI models circumvented controls, reached the internet and compromised parts of OpenAI’s research infrastructure and Hugging Face’s systems.

Anthropic separately disclosed three cases in which Claude models compromised real organizations’ infrastructure after internet access was mistakenly left available during simulated exercises.

Those cases have sharpened concern that cybersecurity is entering a period in which AI systems can not only assist human attackers but also operate with greater speed, persistence and autonomy.

Several signatories are also developing AI tools for cyber defense. OpenAI has expanded Daybreak, Anthropic is giving selected organizations access to Mythos through Project Glasswing, and Microsoft has placed Project Perception in public preview.

The companies said the response cannot rest with any single vendor or government, arguing that the speed of AI development now requires security coordination across industries and borders.