OpenAI will provide $1 billion in subsidized access to its cybersecurity models over the next six months. The program is designed to help organizations defend against AI-enabled attacks and software flaws that attackers could exploit before developers release fixes.

The initiative arrives as critical infrastructure and open-source software have become recurring targets in security incidents, and defenders have warned that AI tooling can shorten the time between a vulnerability's discovery and its exploitation.

The support covers discounted access to Daybreak, OpenAI's cybersecurity product, along with technical assistance and cybersecurity training. According to OpenAI, about 2,000 organizations already use the service.

OpenAI Expands Daybreak Cybersecurity Program

Daybreak offers two service levels. Blue uses OpenAI's standard models for routine defensive work, while Red gives approved organizations access to specialized cyber models for more sensitive security tasks.

The program's priority users include water systems, electricity providers, state and local governments, community banks, nonprofits, and open-source software maintainers. Crypto exchanges, custodiansians, and blockchain networks are not directly named on the priority list.

Open-source crypto projects may still qualify, however, because many rely on small development teams. Bitcoin Core, Ethereum clients, and DeFi libraries support systems that protect large amounts of digital assets.

The announcement follows recent security problems across crypto software. A flaw in BTCPay Server exposed credentials tied to Lightning nodes in August, and attackers used the weakness to steal funds before developers issued a fix. Core Lightning also advised operators to disconnect systems after AI-generated vulnerability reports revealed several real flaws. Coldcard later released new firmware after a major theft and said frontier AI models helped detect unrelated bugs.

Astra Raises New Security Concerns

OpenAI also disclosed that its Astra model can identify unknown software flaws and turn them into working attacks with limited human guidance — without step-by-step direction. That capability has heightened concern over how advanced AI tools could reshape cyber risk, and it feeds into a broader debate in the security community over how much offensive capability AI labs should disclose or restrict.

More than three dozen companies, including Coinbase, Block, BitGo, Blockstream, and ARK Invest, subsequently signed an open letter to AI labs asking for earlier access to powerful models for defensive security work. The companies argued that developers responsible for major crypto systems often receive weaker tools than attackers, and their request focused on giving security teams better access to advanced AI models before criminals can use similar systems.

Eligible organizations can apply through the Daybreak website. OpenAI also plans to expand the program to partner countries in the coming weeks, a rollout that will test whether defensive access can help organizations respond to threats faster. How eligibility is defined for crypto firms and open-source maintainers, and whether subsidized access measurably shortens patch timelines, will be among the details to watch as the program scales.

Source: CoinCentral