Protecting power grids from cyberattacks that could be orchestrated by autonomous artificial intelligence systems was at the center of talks between OpenAI CEO Sam Altman and energy company executives.
Meeting leading energy company executives at the Edison Electric Institute’s annual meeting in Colorado Springs, Altman reviewed efforts to prevent advanced AI models from being misused in ways that could threaten the security of power grids.
The Daybreak initiative takes center stage
The Daybreak cybersecurity initiative, which aims to address vulnerabilities in critical systems, also featured prominently in the talks. The participants discussed opportunities for cooperation to use AI to strengthen the cyber defenses of energy companies.
OpenAI said on July 21 that some AI models tested to measure their cybersecurity capabilities had caused a security breach in the infrastructure of Hugging Face, an AI development platform.
The company described the incident, which involved AI agents capable of bypassing security systems without human intervention and acting in coordination, as a warning about the potential risks of these technologies.