The OpenAI Foundation is committing $130 million in grants to build what it calls "AI resilience"—a layered defense system designed to help society safely harness the rapid advances in artificial intelligence. The initiative spans research, infrastructure, and industry coordination, with early funding already flowing to organizations working in four critical areas: biosecurity, cybersecurity, AI model safety, and the impact of AI on young people.
The foundation's approach draws a powerful parallel to how humanity has managed transformative technologies in the past. When Thomas Edison's Pearl Street Station illuminated Manhattan in 1882, electricity seemed miraculous—until it electrocuted workers and sparked fires in cities across America. Rather than abandon the technology, societies built resilience layer by layer: insulated wires, circuit breakers, independent testing bodies like Underwriters Laboratories, and national electrical codes. Over time, electricity became so safe that a child can flip a light switch without fear. The OpenAI Foundation sees AI following the same trajectory, but compressing centuries of innovation into years.
The stakes are as high as the speed is fast. AI is already lowering barriers to starting a business, expanding access to education, and accelerating scientific discovery in fields like medicine. Yet those same capabilities create mirror-image risks. Systems designed to help young people learn could enable harmful behavior. Tools that accelerate biological research could lower the barrier to creating dangerous pathogens. AI's growing ability to write code could, in malicious hands, threaten critical infrastructure. The technology that upends industries creates winners and leaves workers displaced.
The foundation's initial strategy departs from earlier thinking within OpenAI about how to ensure AI benefits humanity. The early team believed the critical challenge was solving "technical alignment"—keeping AI systems under human control at the technical level. That remains essential work, but the foundation now recognizes it's only one piece of a much larger puzzle. As AI spreads across sectors and borders, society needs independent research institutions, public infrastructure, agreed-upon industry standards, and entirely new fields of expertise built around managing AI's risks.
The four focus areas reveal where the foundation sees both immediate danger and opportunity to act. Bio-resilience work aims to prevent engineered pandemics enabled by AI-accelerated pathogen research. Cyber-resilience efforts seek to protect critical systems from AI-powered attacks. AI model safety research aims to deepen humanity's control over the systems we create. And work on AI's impact on young people is meant to ensure the technology becomes a positive force rather than a source of harm during formative years.
The foundation has already begun distributing money, with more grants coming. Earlier this year, it announced a separate $100 million for life sciences and disease research, betting that AI could help prevent and treat conditions like Alzheimer's. A second program focuses on jobs and economic futures, grappling with what work and prosperity will mean as AI transforms labor markets.
The parallels to electricity are instructive: no single solution protected society from electrical risk. Instead, resilience emerged from many actors—engineers, regulators, standards bodies, and public institutions—working in concert over decades. The OpenAI Foundation is betting that AI resilience will require the same ecosystem approach, built urgently and on purpose, rather than learned through repeated crises.