AI Safety Laws Needed, Anthropic and OpenAI Tell UK Parliament
Anthropic representative Liam Booth-Smith told a UK Parliament hearing that voluntary pledges are no longer enough to ensure AI safety and that new laws are needed. His comments deepened a public disagreement with Google DeepMind and Meta, which declined to support new legislation.
The Business, Innovation, Science and Trade Committee, a cross-party parliamentary committee, held an evidence session on Tuesday attended by representatives of four leading AI companies, four former employees who became whistleblowers, and senior UK civil servants.
Anthropic Calls for Binding AI Safety Rules
“Rules should be binding,” Liam Booth-Smith, head of Europe at Anthropic, told MPs when asked whether AI safety required a legal framework. He said the company had “consistently believed that businesses need binding rules and that voluntary efforts alone are not enough.”
Tom Duff Gordon, vice president of OpenAI, was more cautious but also supported legal frameworks and international treaties. “There should be some elements that should be mandatory,” Duff Gordon said, including incident reporting. “I don’t think it’s a good long-term plan to base everything on a voluntary approach to frontier risk,” he added.
Google DeepMind and Meta Resist New AI Laws
Rival companies did not take the same position. When Google DeepMind’s Owen Larter was asked directly whether he supported a statutory framework, he said: “Focusing on standards is what we need to do.”
Meta’s Rob Sharman pointed to the UK’s existing regulators, telling MPs: “It’s not a blank slate.”
UK Hearing Exposes Divisions Over AI Regulation
The hearing placed the AI industry’s conflicting positions on regulation on the UK Parliament’s record. Liam Byrne, chairman of the task force, said the session was intended to highlight disagreements in jurisdictions outside the companies’ primary markets.
Byrne told WIRED after the hearing that he created the commission to “highlight the deep differences of opinion among some of the wealthiest companies on the planet and show that we cannot rely on voluntary self-regulation.”
MI5 Chief Warns of AI and Biological Weapons Risks
The parliamentary hearing took place just a few hundred meters from an annual threat update in which MI5 chief Ken McCallum warned about the potential misuse of AI in biological weapons.
“While it hasn’t happened yet, the misuse case I’d like to highlight the most is the potential for AI to change the stakes for existing biological weapons and even new biological weapons,” McCallum said.
Source: www.wired.com