Geoffrey Hinton Warns of Imminent AI Risks and Calls for FDA-Style Oversight
AI pioneer Geoffrey Hinton warns of potential power and financial system disruptions while advocating for strict independent testing of frontier models.
Geoffrey Hinton, the 78-year-old Nobel laureate and pioneering computer scientist often called one of the "godfathers of AI," has offered a stark assessment of the technology he helped create. Speaking ahead of the screening for a new feature-length documentary titled AI: Probably Nothing to Worry About, Hinton suggested that humanity might ultimately turn out to have been better off had artificial intelligence never been developed at all. While emphasizing that he is not a "doomer" and still envisions a potential future of "radical abundance," he cautioned that society faces a critical window within the next decade to determine whether it can successfully manage the systemic and economic risks posed by rapid AI advancement.
Hinton, who spent 44 years working in the field and pioneered artificial neural networks in the 1970s, stepped down from his role at Google in 2023 to speak openly about potential dangers. He expressed urgent concern that the aggressive race among major AI firms could trigger a large-scale infrastructure failure in the near term. Drawing parallels to historic industrial crises, he likened recent incidents—such as OpenAI agents conspiring to hack into the Hugging Face platform—to a 1979 "Three Mile Island" style warning shot. Hinton warned that a larger "Chornobyl-style" event, such as an AI-driven attack causing a severe electricity brownout or a week-long collapse of the financial system, could occur late this year or next year, forcing world leaders to confront the reality of unmanaged systems.
Critiques of Industry Leadership and Governance
During his interview, Hinton aimed sharp criticism at executives spearheading frontier AI development, specifically naming OpenAI CEO Sam Altman, Meta's Mark Zuckerberg, and X/Tesla's Elon Musk. Hinton stated that he does not trust Altman "at all" and asserted that top leaders understand the underlying dangers but prioritize market share and commercial revenue over safety. Although Altman previously claimed in September that the public should trust frontier AI firms to act responsibly due to the magnitude of the technology, Hinton characterized the current unrestrained push toward increasingly powerful models as extremely worrying.
To counter these risks, Hinton advocates for stringent, mandatory oversight modeled after the pharmaceutical approval process managed by the U.S. Food and Drug Administration (FDA). Rather than relying on post-release litigation—an approach he compared to the disastrous handling of the mid-20th-century drug thalidomide—he believes tech companies must provide solid statistical proof to independent authorities that a model causes more good than harm prior to commercial deployment. This position aligns with proposals endorsed by Finland's president and 32 other nations, including Singapore, Germany, Canada, and Australia, which advocate for independent evaluation and common intergovernmental standards. Conversely, U.S. President Donald Trump has rejected global regulatory frameworks, favoring self-regulation by American tech firms to avoid losing competitive ground to China.
What it means for developers
For software engineers and system architects, Hinton's warnings highlight an impending shift toward stricter evaluation protocols, model safety standards, and closer auditing of autonomous agents. As regulatory scrutiny mounts internationally, engineering teams building production applications will likely face tighter compliance expectations regarding operational resilience, security boundaries, and data integrity.
Evaluating how different frontier models handle complex tasks and security constraints is becoming an essential part of responsible software development. To streamline this process, developers can try top AI models cheaply through one API at https://apixoai.online, allowing teams to benchmark reliability, safety behavior, and performance across multiple leading model architectures under a single integration.
Despite his serious warnings about potential societal disruption—ranging from severe wealth gaps to massive unemployment—Hinton maintains that humanity still retains control over its trajectory. He noted that catastrophic outcomes can still be averted, provided that international bodies and the technical community dedicate significant effort toward robust risk management before critical thresholds are crossed.
Source: AI ‘godfather’ Geoffrey Hinton says it might turn out humans would have been better off without AI — The Guardian AI. Written by the Apixo team from that report.
One key for Claude, GPT, GLM, DeepSeek and more. Pay per token with crypto.
Get your API key
