OpenAI and Anthropic ask UK lawmakers for regulation

AI company leaders, whistleblowers and security experts appeared at a UK parliamentary session where the labs said they want to be regulated.

At a UK parliamentary hearing, representatives of OpenAI and Anthropic said voluntary safety commitments are no longer enough and that binding rules for frontier AI should be introduced. Google DeepMind and Meta declined to support new legislation.

The hearing

The House of Commons Business, Innovation, Science and Trade Committee, a cross-party group of MPs, held an evidence session with four frontier AI companies, four former employees of those companies who have become whistleblowers, and senior British civil servants.

Anthropic's head of Europe, Liam Booth-Smith, told the committee that the rules should be binding and that the company has consistently believed voluntary commitments are insufficient. OpenAI vice president Tom Duff Gordon was more cautious but still backed a statutory framework and an international convention, saying some elements should be mandatory, starting with incident reporting.

Google DeepMind's Owen Larter, asked directly about a statutory framework, said the company would focus on standards. Meta's Rob Sherman pointed to the role of Britain's existing regulators. Wired notes that, on the same day nearby, the head of a British intelligence agency warned that AI could be used to help create new biological weapons.

Why it matters for builders

For most developers, frontier-model regulation affects you indirectly through the providers you use. If binding rules arrive, expect more documentation from model vendors, formal incident reporting, and possibly usage restrictions in sensitive areas. Some obligations could flow down to companies that deploy models in products.

Incident reporting is the piece most likely to touch builders first. It is worth setting up basic practices now: log serious failures and misuse, keep track of which model versions you use, and have a simple process for reviewing and fixing harmful outputs. These habits are good engineering anyway and will make compliance easier later.

The split between labs is also notable. Companies that support regulation and those that prefer standards may diverge in their policies, which could affect terms of service and availability in different countries.

You can compare the major model APIs in our APIs & Model Platforms category.

Source: Wired. This article is PickAI’s own summary and analysis based on the original reporting.