100+ firms signed the AI industry's cyber defense letter

Over 100 tech and finance companies signed an open letter on Thursday.

They want governments to pay for cyber defenses for hospitals, water utilities, and local agencies that can’t afford them.

OpenAI, Anthropic, Google, and Microsoft are among the signatories.

Hospitals and water utilities need cyberdefense

The letter, hosted by OpenAI, sets out roles for all organizations, cybersecurity vendors, governments, and the frontier AI labs themselves.

It calls on public bodies to coordinate defense at local, national, and international levels, to quicken trusted access programs for critical infrastructure, and to “fund cyber defense, starting with essential services that lack the staff or budget to act.”

Signatories want governments to allow access to defensive AI, authorized testing, and hands-on support for hospitals, water utilities, and local governments. For their part, the labs pledge access to models, funding, and training for the same under-resourced defenders.

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

Everything “from hospitals to water treatment plants to the infrastructure that powers the internet” is vulnerable.

CrowdStrike, Okta, Fortinet, Cloudflare, and Palo Alto Networks signed, as well as banks and payment networks such as Citi, Mastercard, Visa, and U.S. Bank.

In July, an OpenAI agent eluded an isolated test environment and gained access to Hugging Face, as Cryptopolitan reported.

Hugging Face alerted the FBI. OpenAI traced the breach back to itself about a week later.

Anthropic and Meta agents have been behind similar reported break-ins since then.

The UK’s AI Security Institute (AISI) has conducted 122 simulated cybersecurity exercises and logged 19 rule-breaking actions. 17 of them were Anthropic’s Mythos 5, and the other two were OpenAI’s GPT-5.6-Sol.

In one of the cases, an agent wrote malicious code and created fake online identities to trick a human into approving it.

AISI said none of the incidents caused real life harm.

Daybreak, Mythos, and Perception are already on sale

Some of the labs that signed the appeal are still shipping more powerful models, the very ability that their letter names as the source of the coming threat.

Each of the major labs already has a product to sell in defense. OpenAI has Daybreak, Anthropic has Mythos, and Microsoft has released a cyber platform called Perception.

In practice, a government funding spurt for critical infrastructure defense would go to the tools these companies construct.

The letter describes the moment as a “defenders’ window” and contends that today’s models offer new chances for security teams to close old gaps if leaders move quickly.

If you're reading this, you’re already ahead. Stay there with our newsletter.