OpenAI, Anthropic, Big Tech urge cyber ‘defensive surge’ on AI hacks

This letter is a marker that the leading AI providers now see AI‑enabled cyber offense as a systemic risk that threatens their own business models and social license. When OpenAI, Anthropic, Microsoft, Alphabet and Amazon all sign the same warning, it signals that the industry expects adversaries to weaponize large language models and agents for everything from phishing and exploit discovery to automated lateral movement.

Strategically, the call for a “defensive surge” is less about one policy proposal and more about reframing cybersecurity as a race against AI‑accelerated attackers. The big clouds are implicitly arguing that only a coordinated upgrade of tooling, standards and talent will keep pace, and that they intend to sell a lot of that tooling. It also helps them get ahead of the regulatory curve by positioning themselves as responsible actors rather than purely profit‑driven vendors of risky technology.

For the race to AGI, this move will not slow capability work directly, but it will push more capital and attention into defensive automation, secure‑by‑default infrastructure and joint public‑private response mechanisms. Over time, those shifts could make high‑end AI development politically more sustainable, because governments and critical industries will feel less exposed even as models grow more capable.