OpenAI Commits $1B to Cybersecurity After AI Finds Zero-Day Flaws

OpenAI is subsidizing access to its Daybreak cybersecurity platform after revealing that its Astra model can independently identify previously unknown software vulnerabilities and develop functional exploits from them.

The company said Thursday that it will offer $1 billion worth of subsidized access to its cybersecurity models over the next six months. The initiative is designed to help organizations strengthen their defenses against AI-powered attacks, including potential zero-day exploits generated by future systems such as Astra.

Zero-day exploits target software vulnerabilities that have not yet been publicly identified or patched, giving attackers an opportunity to exploit them before developers can respond.

The $1 billion commitment is not structured as a traditional fund or grant program. Instead, it will provide organizations with discounted access to Daybreak, the cybersecurity product OpenAI introduced earlier this year, along with training and technical assistance.

Daybreak is available in two versions. Blue uses OpenAI’s standard models for common defensive cybersecurity tasks, while Red provides approved organizations with specialized models designed for more sensitive security operations. OpenAI said roughly 2,000 organizations are already using the platform.

The program will prioritize organizations operating water and wastewater infrastructure, electric grids, state and local governments, community and regional banks, nonprofits and open-source projects. Crypto exchanges, custodians and blockchain protocols are not specifically identified in the announcement.

However, much of the open-source software underpinning the crypto industry falls within the broader category of systems that could benefit from stronger cybersecurity resources. Bitcoin Core, Ethereum clients and the software libraries supporting DeFi protocols are often maintained by relatively small teams and volunteer contributors, despite securing billions of dollars in digital assets.

Astra Raises the Stakes for Cybersecurity

The urgency surrounding OpenAI’s initiative follows the company’s disclosure on Monday about Astra. The new model can reportedly discover previously unknown software vulnerabilities and convert them into working attacks without requiring a human to guide each stage of the process.

That capability represents a significant shift in cybersecurity, as sophisticated vulnerability discovery and exploitation has traditionally required skilled security researchers or hacking teams.

Organizations seeking access to the subsidized program can apply through Daybreak’s website. OpenAI also said it plans to expand availability to partner countries in the coming weeks.

The announcement arrives during a challenging period for crypto applications and blockchain infrastructure, where several recent incidents have highlighted vulnerabilities in widely used open-source software.

In August, a security flaw in BTCPay Server, an open-source platform used by merchants to accept Bitcoin payments, exposed credentials associated with Lightning nodes. Attackers exploited the vulnerability and drained funds before a patch became available.

Core Lightning, one of the primary software implementations used to operate Bitcoin’s Lightning Network, also warned operators last week to disconnect their systems following a series of AI-generated vulnerability reports that identified multiple genuine flaws.

Coldcard, a hardware wallet designed for offline Bitcoin storage, separately released updated firmware following a $114 million theft. The company said advanced AI models had helped identify bugs unrelated to the vulnerability behind the original incident.

Crypto Industry Seeks Stronger AI Security Tools

The incidents have fueled growing concern among crypto companies about the widening gap between the capabilities available to attackers and those accessible to developers responsible for securing critical infrastructure.

More than three dozen companies, including Coinbase, Block, BitGo, Blockstream and ARK Invest, recently signed an open letter calling on AI laboratories to provide security professionals with early access to their most capable models.

The companies said AI labs were carefully evaluating which organizations could access their strongest systems. At the same time, attackers could potentially obtain or exploit those capabilities through other means, leaving teams responsible for maintaining critical Bitcoin and blockchain infrastructure with less advanced tools than those available to their adversaries.

OpenAI’s $1 billion Daybreak initiative is therefore aimed at strengthening the defensive side of that equation, particularly as AI systems become increasingly capable of discovering and exploiting software vulnerabilities.

About The Author