TL;DR

Anthropic has hired Charlesbye, a lobbying firm managed by Boris Johnson’s former communications director Lee Cain, Bloomberg reported on 10 November. The contract shows up in the UK lobbyist registry for the most recent quarter, the first time Anthropic has appeared there. The hire comes as the Claude developer prepares its stock market listing and the UK has yet to decide how to legislate on AI.

What the filing shows

The UK registry lists clients who pay consultant lobbyists to contact ministers or permanent secretaries. Charlesbye markets itself on public affairs, crisis management and corporate strategy. Cain declined to comment to Bloomberg, and Anthropic would not discuss the contract or how many former officials it employs.

An Anthropic spokesperson said the company backs the government’s efforts “to understand and manage the risks and benefits from power AI”.

A bench of Westminster veterans

Bloomberg says Anthropic has gone further than its rivals in recruiting from Downing Street. Former prime minister Rishi Sunak and his chief of staff Liam Booth-Smith have joined, and Matt Clifford, who advised the next prime minister on AI, followed in September. The report counts at least nine other former officials from the Treasury, the tech department and Cabinet departments, based on LinkedIn profiles. OpenAI, by comparison, hired former chancellor George Osborne in 2025 to lead its international work.

Regulation still unsettled

The timing reflects an open question in UK policy. Labour’s manifesto promised “binding regulation” of AI model developers, but no such law has passed. AI minister Kanishka Narayan, appointed to the cabinet in July, has pledged that the government would “legislate wherever necessary”. We reported in September that planned AI safety legislation had fallen away.

What it means

Anthropic now counts a former prime minister, his chief of staff and, through Charlesbye, a former head of Downing Street communications among its advisers, while the UK is still deciding what AI law it wants.

In our view, the register entry is the useful part. It puts the relationship on the public record, so journalists and MPs can see which AI companies are paying for access to ministers. That record will matter more as the government decides whether binding rules for model developers come back onto the agenda.