World may have just one year to rein in AI: Geoffrey Hinton

Nobel-winning computer scientist Geoffrey Hinton has warned that governments may have only about a year to develop effective safeguards for increasingly powerful artificial intelligence systems before controlling them becomes significantly harder.

Speaking after a closed-door meeting with US lawmakers in Washington, Hinton was asked how much time policymakers have to act. “Maybe one year, but not much longer,” he said, according to reports.

Hinton, widely known as the “Godfather of AI”, has urged a slowdown in AI development alongside stronger safety measures.

He cited the recent incident involving OpenAI’s AI agents and Hugging Face as a warning sign. During an internal cybersecurity evaluation in July, OpenAI models circumvented controls intended to isolate them from the internet, exploited vulnerabilities and accessed parts of Hugging Face’s infrastructure.

The agents also developed unauthorized ways to communicate and share information among themselves, according to OpenAI.

Hinton’s concern is also driven by the rapidly shrinking timeline for artificial general intelligence (AGI)—systems capable of performing a broad range of tasks at human or potentially higher levels.

He said estimates that once put AGI decades away have increasingly moved toward a timeframe of just a few years.

The Hugging Face incident has intensified concerns about the ability of advanced AI agents to operate beyond their intended constraints. Hugging Face’s own technical reconstruction found that an autonomous agent carried out thousands of automated actions during the intrusion, underscoring the growing capabilities of frontier AI systems.