OpenAI commits $1B in AI credits for cyber defense

OpenAI has committed $1 billion in AI credits to cybersecurity teams defending critical infrastructure, community banks, and open-source projects that can't afford advanced AI tooling on their own. The Daybreak for Frontline Defenders program runs over six months and pairs the credits with hands-on training for security staff at eligible organizations.

Eligible groups include critical infrastructure operators, community banks, nonprofits, open-source maintainers, and state, local, tribal, and territorial government defenders. A pilot through the Multi-State Information Sharing and Analysis Center starts with water utilities and other public-sector systems before expanding further.

Participants get access to GPT-5.6 Sol for defensive security workflows and GPT-5.6 Cyber for offensive testing, with a promised future upgrade to OpenAI's Astra model on a timeline the company hasn't specified. Organizations can apply directly through OpenAI.

OpenAI president Greg Brockman framed the program as a response to a widening gap between attackers and defenders: "We might be heading to a world where critical infrastructure outages are just a way of life," he said, arguing that under-resourced teams need the same AI capabilities attackers already have.

For an open-source maintainer or an under-resourced security group without budget for enterprise AI tooling, Daybreak is one of the few paths to that access without paying commercial rates, though acceptance criteria and actual capacity limits aren't detailed yet.