OpenAI Launches $1 Billion Initiative to Strengthen Cyber Defense for Utilities

In a significant announcement made at OpenAI’s headquarters during a summit with 300 security professionals and CISOs from Fortune 1000 companies, OpenAI President Greg Brockman introduced the Daybreak for Frontline Defenders initiative. This global program is designed to empower frontline defenders with advanced cyber AI tools to safeguard essential services both in the United States and worldwide.

The initiative marks a $1 billion commitment to broaden access to Daybreak cyber models, which include subsidized training, technical support, and collaborations locally and globally. Central to Daybreak is its defensive model, consisting of frontier models, the Codex harness—an execution engine connecting AI models to user systems—and Codex Security, a tool for identifying and addressing vulnerabilities in connected codebases and workflows.

Brockman illustrated the efficacy of this initiative by sharing his experience using Codex to assess and rectify vulnerabilities on his personal website. “I’m not an expert, but Codex is,” he stated. “It fixed all the issues I found, and I felt well-protected.”

Enhancing Cyber Defense for Essential Infrastructure

A pivotal element of this initiative is the Daybreak for America project, which focuses on shielding smaller water and electrical service providers, local governments, and banks. To kick off this effort, OpenAI will initiate a pilot program in collaboration with the Multi-State Information Sharing and Analysis Center (MS-ISAC), targeting training and support for cyber defenders in state, local, tribal, and territorial entities, starting with public-sector defenders.

OpenAI aims to combine Daybreak access with tailored training and hands-on assistance for an initial cohort of public sector defenders. This approach will enable them to prioritize vulnerabilities, strategize remediations, and develop a methodical approach for sustained improvement.

Brockman pointed out, “We might be heading to a world where outages in critical infrastructure are common. We have a window to prevent that, but action is necessary.”

Expanding on this initiative, OpenAI held its second meeting with utility firms, aimed at equipping them with tools to strengthen their systems. Participants from 40 states and the District of Columbia serve a combined population of over half the U.S.

Recently, OpenAI's CEO Sam Altman highlighted the organization’s role in partnering with cybersecurity professionals to protect essential services in North Carolina. This effort is part of a larger coalition of over 150 organizations from various sectors—including cybersecurity, technology, and finance—advocating for a unified response to the cybersecurity challenges facing critical infrastructure.

Brockman warned about the urgency of addressing vulnerabilities, stating, “If a hole exists, the risk of them being exploited has increased significantly. The once acceptable best practices can no longer guarantee safety.”

In response to recent cyberattacks on U.S. water systems, OpenAI has stepped forward to offer impacted states and utilities up to $1 million in free API credits, Daybreak access, and technical assistance. Affected teams are utilizing this support to inspect code, validate configurations, develop patches, and confirm fixes—all while preserving essential service continuity.

Eligible entities, including state and local governments, critical infrastructure operators, nonprofits, and open-source maintainers, can visit the Daybreak website for information regarding access, technical support, training, and other cybersecurity assistance.

Expert Insights on Implementation Challenges

Though utility experts commend the initiative, they express concerns regarding its practical implementation. Gus Serino, an industrial control security expert and president of I&C Secure, noted, “I commend this effort, but AI can't truly assess operational technology without a detailed human overview.” He stressed the utility sector's need for hardware and software identification, procurement, installation, configuration, and maintenance—tasks that still require human participation despite AI support.

Patrick Miller, CEO of Ampyx Cyber, echoed these sentiments, cautioning against overwhelming utilities with tools that are difficult to implement promptly. “Identifying more vulnerabilities is the easy part, and that’s already surpassing what they can manage effectively. While the defensive strategies are promising, they need to be rigorously tested before deployment in operational technology environments,” he advised.