US Vice President JD Vance on 30 September warned artificial intelligence companies against seeking sweeping liability protections, saying developers of powerful frontier models must own responsibility for safety rather than lobby Washington for a new regulatory umbrella. The remarks followed a White House meeting with technology industry leaders.

No Section 230-Style Shields for AI

Vance was direct on the question of legal protection. 'What I think would be a huge mistake is to give, for example, Section 230 protections to some of the new technology companies,' he said. 'That actually shields them from liability rather than making them more responsible.' The reference to Section 230 — the provision that has long protected internet platforms from lawsuits over user content — signals the administration's resistance to extending similar cover to AI developers whose models generate content and take consequential actions autonomously.

Against 'Cartel-Like' Behaviour Among Top AI Labs

Vance also cautioned against collective action among the handful of dominant AI developers. 'We don't want the four or five biggest model companies… to join in a sort of unified cartel against the American consumer,' he said, adding that President Donald Trump's administration wanted those companies competing against one another. He acknowledged that leaders of several major AI firms were 'very smart, earnest people' but argued that goodwill did not substitute for market competition. The administration pointed to the Federal Trade Commission and Department of Justice as the appropriate tools for antitrust enforcement, and stressed the need for new entrants to be able to compete.

Existing Law Over New Agencies

Rather than a dedicated AI regulator modelled on the Food and Drug Administration or Federal Aviation Administration, Vance said existing product liability principles offered a more workable framework. 'Most bureaucrats just know way less about this than the people who are actually building these products,' he said. He added that regulations forged in a climate of panic tended to produce poor outcomes, citing COVID-19 era policy as a cautionary example: 'I don't think anybody regulates well in the middle of a panic. COVID should have taught us that.'

Cyber Risk and 'Super Intelligence'

Vance identified cyber capabilities as among the most tangible near-term risks from frontier models, noting that some systems had a 'remarkable ability' to help hackers penetrate computer systems. 'The best weapon against cyber offense is good cyber defense,' he said. On broader public anxieties, he rejected the idea that advanced AI — which the Trump administration has begun calling 'super intelligence' — would inevitably escape human control. 'They think the terminators are gonna come and take over everything… And that's just not gonna happen,' he said, while still acknowledging risk if companies fail to exercise sufficient care. 'If you're building a terrible, terrible thing, then stop and build something good. Don't come to the government and regulate,' he added.

What This Means for Global AI Policy

The remarks crystallise the Trump administration's emerging posture: strong on competition and accountability, sceptical of large new regulatory structures, and resistant to liability immunity that could reduce incentives for safety. This comes amid a broader global debate — in the European Union, the United Kingdom, and now India — over how to govern increasingly capable AI systems. The administration's emphasis on antitrust and product liability rather than sector-specific oversight could influence how trading partners and multinational tech companies calibrate their own compliance strategies. Further policy detail is expected as the administration continues its engagement with the technology industry.