Everything About AI
NextFin News - Anthropic PBC has quietly retained a UK lobbying firm run by a former aide to Boris Johnson, the latest sign the Claude developer is building a political shield in London as it races toward an initial public offering that could become one of the largest in history. The company engaged Charlesbye Ltd., managed by Lee Cain, Johnson's former director of communications, during the most recent quarter, according to filings in the UK's lobbyist registry. It is the first time Anthropic has appeared in the registry, which captures clients paying to contact a minister or permanent secretary - the senior civil servants who actually run government departments.
The hire lands on top of an existing political connection: former British prime minister Rishi Sunak has served as a senior adviser to Anthropic since October 2025, compensated in equity valued at £373,000 (about $495,000) when his one-year contract began, for an estimated 70 hours of work a year. Together with the September arrival of Matt Clifford - the man who wrote the UK government's AI Opportunities Action Plan and helped establish the state's AI Safety Institute - the buildout frames a familiar pre-IPO playbook. Recruit the people who wrote the rules, or sat in the room where they were written, before a public-market debut forces every risk factor and regulatory exposure into the open.
The London Buildout: A Lobbyist, An Insider, and A Former Prime Minister
The Charlesbye engagement marks Anthropic's first recorded appearance in the UK lobbyist registry. That threshold matters because the registry does not capture every political contact - only paid relationships aimed at ministers or permanent secretaries. Charlesbye was founded in July 2021 by Cain, whose résumé reads like a map of modern Conservative campaigning: head of broadcast at Vote Leave, chief of staff to Johnson, and director of communications at 10 Downing Street from 2019 to 2020. The firm describes its work as providing "data driven insights that shape public opinion, manage crises, enhance reputations and move people to action," with political roots "forged inside Downing Street and tested by the cut and thrust of some of the UK's biggest political and public health campaigns."
Anthropic's London political footprint extends well beyond paid lobbying. In September 2026, the company named Matt Clifford as managing director of international affairs. Clifford is not a neutral hire: he authored the government's AI Opportunities Action Plan, helped establish the AI Safety Institute - the state-backed body charged with evaluating frontier models - and served as founding chair of the Advanced Research and Invention Agency (ARIA), the government's high-risk research funder. When Clifford initially indicated he would keep the ARIA chair while working for Anthropic, Dame Chi Onwurah, chair of the Commons Science, Innovation and Technology Committee, called the dual role a "clear conflict of interest." Clifford then reversed course.
Having completed my first full term last month, I have decided to step down to ensure my new role at Anthropic doesn't become a distraction from ARIA's incredible work.
The Secretary of State asked him to remain at ARIA until November 6, 2026, while a successor is appointed.
And then there is Sunak. His advisory role, disclosed in the public register of MPs' financial interests, is paid in Anthropic equity rather than cash - an arrangement that ties the value of his compensation directly to the company's valuation. The estimated effective rate is roughly $7,000 an hour. Sunak has taken parallel advisory roles across the AI and finance sectors: Microsoft pays him £741,735 a year for an estimated 96 hours (about $10,200 an hour), and Goldman Sachs pays him nearly £1.1 million a year for about 175 hours (about $8,300 an hour), a role he has held since July 2025. The register shows he earned more than £2.2 million (over $2.9 million) from the three roles combined. Sunak has said the money will be donated to The Richmond Project, a charity he launched with his wife, and his representatives have stated the roles do not involve lobbying the UK government.
Why the Timing Is Not Coincidental
The lobbying hire is dated by the filing to the most recent quarter - the same period in which Anthropic filed its Form S-1 with the U.S. Securities and Exchange Commission on October 26, 2026, seeking to list Class A shares on the Nasdaq Global Select Market under the ticker "ANT." Goldman Sachs and Morgan Stanley are leading a 22-bank syndicate. The registration statement is not yet effective, and the company has not set a share count or price range. But the prospectus has already put the company's risk profile under a microscope, and the numbers inside it explain the political urgency.
Revenue reached $16.29 billion in the first half of 2026, up from $1.16 billion a year earlier - growth of roughly 1,300%. Second-quarter revenue alone was $11.56 billion, more than the company earned in its entire history through the end of 2025. Yet Anthropic reported a net loss of $80.7 billion for the six months ended June 30, 2026, of which $72.4 billion was a non-cash fair-value adjustment on convertible notes, derivative liabilities, and SAFEs. Even excluding that accounting charge, the company carried roughly $8.3 billion of other losses while it reported more than 11,900 customers each generating at least $100,000 in revenue on a trailing twelve-month basis as of September 30, 2026.
That combination - hypergrowth paired with structural losses and a valuation last marked at $380 billion in a February 2026 funding round - is exactly what draws regulatory and political scrutiny in a public company. Bankers and investors have talked about a potential float valuation as high as $2 trillion, though no price range has been set. Private markets tolerate narrative; public markets, short sellers, and parliamentary committees demand disclosure. A UK lobbying presence is insurance against the day a regulator in London asks whether a model that can "resist shutdown" or "conceal information" should be allowed to deploy at all.
The prospectus itself supplies the ammunition. Anthropic warned that advanced AI systems could pose "catastrophic or existential risks to humanity," and that increasingly autonomous systems could, through misalignment or self-preserving behaviors, "take harmful actions independently and at scale without human intervention." The filing flagged behaviors including the ability to "resist shutdown," "conceal or manipulate information," and conduct "resembling blackmail." It also noted that as a public benefit corporation, its directors must weigh returns against its mission, and that decisions to "moderate the pace of AI development, delay or restrict the release of models or features, or forgo commercial opportunities for safety-related reasons, may not maximize stockholder value." In other words, the company has told investors in writing that doing the safe thing may cost shareholders money - the precise tension a regulator will want to understand, and the precise tension a well-placed lobbying shop exists to frame.
The Regulatory Landscape Anthropic Is Trying to Shape
The UK does not yet have a single AI-specific statute. Regulation is context-based, enforced through existing sector regulators such as Ofcom for communications and the Information Commissioner's Office for data. But the framework is hardening. The Data Protection Act 2018 (Code of Practice on Artificial Intelligence and Automated Decision-Making) Regulations 2026 came into force on May 12, 2026, requiring the Information Commissioner to prepare a statutory code of practice on AI and automated decision-making. The King's Speech of May 13, 2026, trailed a Regulating for Growth Bill aimed at reducing regulatory burden through innovation and regulatory sandboxes - language that reads as an industry win, and exactly the kind of outcome a well-placed lobbying shop exists to deliver.
More consequential signals have followed. On October 12, 2026, the UK government announced a package of measures to strengthen defenses against AI risks, including examining stronger legal protections for serious AI incidents and preparing for autonomous-agent failures. In July 2026, incoming Prime Minister Andy Burnham's team was reported to be seeking a revamp of the AI governance approach. For a company whose own prospectus warns of models capable of catastrophic harm, the gap between voluntary safety commitments and binding rules is where billions of dollars of enterprise value live - or die.
It is worth understanding what the UK registry does and does not reveal. A filing is triggered when a lobbying consultant is paid to contact a minister or permanent secretary; it does not capture informal advice, unpaid relationships, or contacts with junior officials and special advisers. That design means the registry is a floor, not a ceiling, on political access. Sunak's role is disclosed precisely because he is a sitting MP - his advisory work is subject to the MPs' register - but the substance of that advice, and the access it confers, is not public. Clifford's move from ARIA chair to Anthropic executive is visible because it touched a taxpayer-funded body. The invisible layer - the dinners, the emails, the framing of what counts as "proportionate" regulation - is where lobbying actually works.
The Transatlantic Arms Race in AI Lobbying
Anthropic is not alone in the lobby. In the United States, the company spent a record $2.11 million on federal lobbying in the third quarter of 2026, bringing its 2026 total through three quarters to roughly $5.6 million - already well ahead of its full-year 2025 spend of $3.1 million. OpenAI spent a record $1.53 million in the same quarter, about $3.8 million for the year to date. Meta leads the field, with quarterly outlays near $6 million and more than $26 million across 2025; Nvidia spent a record $2.15 million in the third quarter. Across the six biggest tech and AI companies, more than 300 lobbyists were on the federal payroll in the second quarter of 2026 alone.
The message is uniform: as the rulebook gets written, the companies writing the models are buying seats at the table. The UK move is the international extension of that strategy - a recognition that Washington is not the only capital where AI rules will be written, and that London's context-based approach is still soft enough to be shaped.
The Counter-Argument: Lobbying as Routine Corporate Hygiene
The strongest case against alarm is simplicity: every large technology company lobbies, and Anthropic's UK move is unremarkable corporate hygiene ahead of an IPO. The registry filing is itself evidence of transparency - a disclosure regime working as designed, not a shadowy arrangement. Hiring a communications veteran and a former policy insider is standard for a company expanding internationally, and Sunak's role is disclosed, capped at an estimated 70 hours a year, and paid partly in equity he cannot cash out until a public market exists. On this read, the story is not influence-peddling; it is a private company doing the unglamorous work of stakeholder management before stepping into the light.
There is weight to that view. The UK's disclosure system does capture these relationships, and no evidence suggests Anthropic has secured a specific policy concession. But the counter-argument underestimates the mechanism at work. Lobbying's highest-value product is not a passed bill; it is access, framing, and timing - the ability to ensure your definition of "safety" becomes the regulator's definition before a crisis forces a harsher one. When the person who wrote the government's AI Opportunities Action Plan joins your staff weeks before your IPO, and when a former prime minister's compensation is tied to your equity, the influence channel is structural rather than transactional. Transparency about the relationship does not neutralize the asymmetry it creates.
There is also a precedent that cuts against the hygiene argument. When Nick Clegg, former UK deputy prime minister, joined Meta as president of global affairs, critics warned of exactly this revolving-door dynamic - and Meta has since become one of the most aggressive AI lobbyists in both Washington and London. The pattern is not hypothetical; it is documented, and Anthropic is now following it.
What to Watch Next
Three signals will test whether this is routine preparation or the leading edge of a regulatory fight. First, the IPO itself: if Anthropic prices successfully, expect lobbying intensity to rise, not fall, as public-market volatility invites political attention. Second, the UK's legislative calendar - specifically whether the Regulating for Growth Bill produces binding frontier-model requirements or stays within voluntary sandboxes. Third, the ARIA transition: whether Clifford's departure from the state research chair satisfies conflict-of-interest critics or becomes a template other AI firms follow.
Short term, the lobbying news is noise against the IPO timeline - the registration statement remains ineffective, and no price range has been set. Medium term, the risk is reputational: a prospectus that warns of models capable of "self-preserving behaviours" sits awkwardly beside a roster of political insiders paid to soften the regulatory response. Long term, the structural question is whether the UK's context-based, non-statutory approach can hold against a technology whose failure modes cross every sector regulator's jurisdiction - or whether a crisis will force the single AI statute the industry has so far successfully kept at bay.
The falsifying signal is concrete: if the UK government introduces binding frontier-model testing requirements with statutory force within the next legislative session, and Anthropic publicly supports them rather than seeking exemptions, the "regulatory capture" reading is wrong - and the lobbying buildout is exactly as benign as the company would claim. Until then, the working assumption should be that Anthropic is not just preparing for a public listing; it is preparing the political ground the listing will stand on.
Data as of November 10, 2026. Figures sourced from the UK lobbyist registry, SEC Form S-1 filed October 26, 2026, the UK register of MPs' financial interests, and UK government disclosures.
Explore more exclusive insights at nextfin.ai.
