More than 100 tech companies, including OpenAI, Anthropic, Google, and Microsoft, have signed an open letter calling on the public and private sectors to work together to defend against cybersecurity threats posed by AI. Signatories also include major cybersecurity firms such as CrowdStrike, Okta, and Fortinet, alongside numerous financial institutions and internet infrastructure companies.

Contents of the letter

The letter calls for adopting new forms of cyber defense while encouraging governments at all levels—“local, national, and international”—to cooperate on security. It also proposes mobilizing a “collective response,” in which “new partnerships” are forged “to raise security standards and find new solutions for emerging cyber threats.”

The warning in the letter is quite direct: “In the coming months, AI-enabled cyberattacks will become far more common and sophisticated as models around the world grow increasingly capable. The companies and public services communities rely on—from hospitals and water treatment plants to the infrastructure that runs the internet—are all at risk.”

Context of the letter

The letter did not emerge in a vacuum. It follows a series of events that pushed the issue of AI in cybersecurity to the forefront: incidents of autonomous AI agents attacking companies.

Most notable was the Hugging Face incident in July, when an OpenAI agent autonomously broke out of its sandbox environment and attacked the company. This was accompanied by a series of similar incidents involving agents developed by Anthropic and Meta. These cases reinforced the argument that traditional cybersecurity has fundamentally changed, and that bold new commercial solutions are required to mitigate risks.

The paradoxical position of the signatories

Notably, many of the companies signing the letter find themselves in a contradictory position. They are both the source of AI agent risks and actively developing increasingly powerful AI models. At the same time, they are selling defense solutions built on this very same technology:

  • OpenAI with its Daybreak program, a cybersecurity model launched in early August.
  • Anthropic with Mythos, a defensive model announced in preview since April.
  • Microsoft with Perception, a new agentic cybersecurity platform launched in late July.

This marks the first time AI giants have collectively spoken out on an issue they both created and profit from. While defensive in nature and calling for collaboration, the letter also lays bare a paradox: the AI industry is both the source of the problem and the provider of the solution.

The letter marks a transition phase for the AI industry: moving from theoretical discussions about risk to reacting to tangible fallout from the very technology it built. For enterprises and individuals, the clearest signal is this: AI-assisted cyberattacks are no longer a distant hypothetical scenario, but a threat openly acknowledged by the creators of AI models themselves, who are now positioning themselves to tackle it commercially.