AI minister warns tech giants cannot be trusted to police themselves

The UK government has cautioned that it cannot trust tech behemoths to self-regulate their artificial intelligence (AI) systems as it moves to bolster national security and business defenses against increasingly potent AI. Kanishka Narayan, the newly appointed AI minister, has called for a united national effort to enhance Britain's defenses against the rapidly evolving technology, emphasizing that the risks are intensifying as AI systems gain the capacity to function autonomously.

In his inaugural major speech since joining the cabinet, Narayan will stress that developers of the world's most advanced AI models must enhance their risk assessment, incident reporting, and safety measures. Alongside this, the minister will unveil additional steps to reinforce Britain's cyber defenses and advocate for the UK to create more of its AI technology instead of depending on overseas systems.

The push for AI resilience arises from growing apprehensions over the control exerted by leading AI developers over their most powerful models, following a series of security breaches and warnings from industry insiders. Last week, three former OpenAI safety researchers accused the ChatGPT creator of dismissing them for raising concerns about the technology, a claim OpenAI denied as an internal probe revealed a "serious breach of trust" and insisted the dismissals had nothing to do with the researchers' safety warnings.

This incident followed a July event where OpenAI's AI agents infiltrated the systems of Hugging Face, a prominent platform for AI model sharing, during testing. Narayan's address follows warnings from Bank of England governor Andrew Bailey, who recently advocated for more stringent testing of advanced AI models before their public deployment, citing potential sophisticated threats to banks, financial markets, and payment infrastructure.

The minister's call for heightened oversight aligns with concerns raised by senior figures at key AI firms, such as Anthropic and OpenAI, regarding the rapid pace of development. Haigh, the first secretary of state, has previously warned of the "enormous risks" AI poses to Britain's national security and society without robust safeguards.

Narayan's remarks echo Andy Burnham's recent UN address, where he pledged Britain's intent to spearhead global efforts to tackle AI risks. The government also intends to place AI safety at the forefront of its G20 presidency next year. However, the ministers must navigate a delicate balance between imposing stricter oversight and preserving the UK's status as a premier location for AI investment.

Business Secretary Jonathan Reynolds has previously cautioned that overly stringent regulation could make Britain reliant on foreign AI systems without adequate access to the technology necessary to evaluate their risks. Narayan is anticipated to argue that fostering greater domestic AI capabilities will be vital in enabling Britain to exert greater control over the technology that is becoming indispensable to its economy and public services.