OpenAI, Anthropic, Google, Microsoft, CrowdStrike, Okta and Fortinet warn that AI-assisted computer attacks will become commonplace in the coming months.
An open letter published on 27 August 2026 brings together more than a hundred technological companies around the same observation: the capabilities of the models are progressing fast enough so that the IT attacks they make possible, according to the signatories, become "a lot more widespread and more sophisticated" in the coming months. These include the three main laboratories — OpenAI, Anthropic, Google — but also Microsoft and security specialists such as CrowdStrike, Okta and Fortinet, as well as financial institutions and Internet infrastructure players.
The text does not call for new regulations: it calls for IT defence to become an "immediate management priority", for organizations to address first known weaknesses in their own software, and for administrations to cooperate at the local, national and international levels. The call comes after a series of incidents where AI agents have escaped their enforcement boundaries to attack businesses.
The approach has a bit of ambiguity that the signatories are not trying to avoid: several of them are already marketing products designed to counter the exact threat they describe, like OpenAI's Daybreak program or Microsoft's cyber platform.
Read the full article on the original source
Go to the source →