The Revolving Door Opens: Sunak's Dual Advisory Role Exposes the Policy Arms Race

The LinkedIn update read like every other post-bureaucrat flex. New profile photo. New headline. 'Advisor to Microsoft and Anthropic.' No mention of the 130 billion reasons why those two companies might want a former G7 head of state on speed dial. Gas fees don't lie. Neither do career moves. The former UK Prime Minister didn't just accept a board seat or a ceremonial title. He planted a flag in the middle of the most contested terrain in AI: the space where code meets legislation. And he did it for both sides of the same strategic alliance. Let's be clear about what this is. Sunak is not joining to review model architectures. He's not auditing attention mechanisms or debating the merits of constitutional AI. He's being hired for his Rolodex, his understanding of the regulatory machinery, and the fact that he personally convened the first global AI Safety Summit at Bletchley Park in November 2023. That summit produced the Bletchley Declaration. It also produced a network of contacts that no venture fund can buy. The timing is not accidental. The EU AI Act passed but its implementation details are still being hammered out. The US executive order on AI is being operationalized. The UK's own AI white paper promised a 'pro-innovation' framework but delivered mostly ambiguity. This is the policy window. And windows close. What Sunak brings is the ability to walk into any finance ministry or regulatory body in the Western world and get a meeting within 48 hours. That is a commodity. And in the current AI landscape, it's becoming the most valuable one. Consider the competitive structure. Microsoft has invested roughly $13 billion into OpenAI. It has also invested a similar amount into Anthropic. This dual-track strategy is a hedge. But it creates a coordination problem. How do you support two competing frontier labs without alienating either? The answer, apparently, is to install a former prime minister as a shared policy asset. It's a masterstroke of alignment. Sunak serves both masters because their policy interests currently converge: keep the regulatory burden predictable, keep the compute flowing, and keep the narrative focused on 'safe' AI rather than 'dangerous' AI. Anthropic's calculus is more nuanced. The company has built its brand on safety. It's a Public Benefit Corporation. It talks about 'Constitutional AI' and long-term existential risk. But its business model requires selling Claude to enterprises and governments. Those buyers are skittish. They want assurance that the vendor won't be regulated into oblivion. Having Sunak on the masthead provides that assurance. It signals that the company has a direct line to the people who write the rules. It's not just a technical differentiator. It's a trust anchor. My audit experience tells me to look for the asymmetry. In 2020, I watched a yield aggregator collapse because the team spent more time on Discord than on reentrancy guards. The pattern here is similar but inverted. Sunak's appointment is not about technical competence. It's about structural positioning. The question is whether this positioning creates value or simply extracts it from the public trust. The ethics dimension is where the ledger gets messy. Sunak was in office when the UK's AI regulatory framework was being shaped. He left office. He joined the very companies that are most affected by that framework. This is the classic revolving door. The UK's Advisory Committee on Business Appointments (ACOBA) reviews such appointments, but its recommendations are not legally binding. There is no enforcement mechanism. There is only optics. And the optics are bad. The public already distrusts AI. The 2024 Edelman Trust Barometer showed only about 40% of global respondents trust the technology. When they see a former head of government cashing in on his policy influence, that number doesn't go up. It hardens the narrative that AI is controlled by a small elite that plays both sides of the regulatory game. But here's the contrarian angle. The bulls might be right about the substance. Sunak genuinely pushed for AI safety during his tenure. He hosted the summit. He signed the declaration. He may actually believe that engaging with frontier labs from the inside is more effective than shouting from the outside. There is a coherent argument that his advisory role will make Microsoft and Anthropic more cautious, not less. That his presence will push them toward the 'responsible' end of the spectrum because he has a reputation to protect. I've seen this before. In 2017, I found a critical reentrancy vulnerability in a token contract. The code was beautiful. The intent was fraudulent. The developers had spent weeks polishing the UI and months ignoring the logic. I sent them a patch privately. They never deployed it. They were too busy marketing. Intent is fiction. Code is truth. But policy is neither. Policy is a negotiation between power centers. Sunak's appointment is a data point in that negotiation. It tells us that the AI industry has concluded that technical superiority alone is insufficient. You need regulatory immunity. You need political cover. You need the ability to shape the rules before they are written. This is the new arms race. Not just compute and talent. But access and influence. The ledger will keep score, but the scoreboard is now in Westminster, Brussels, and Washington. What remains unknown is the price of this access. What does Sunak get paid? Stock options? A consultancy fee? And what does he give in return? An introduction here. A quiet word there. The details are hidden in the fine print of corporate governance, far from the transparency that AI governance supposedly demands. The deeper question is whether this accelerates the 'co-regulation' model where companies write their own rules with friendly ex-politicians providing the legitimacy. Or whether it triggers a backlash that results in heavier, more prescriptive regulation. I would bet on the former. The system is designed to absorb critics. Sunak is now inside the system. He will be effective. But his effectiveness will be measured in policy outcomes, not in the safety of the models. And that is the tragedy. We are outsourcing the governance of our most powerful technology to the very people who profit from its unregulated deployment. Minted nothing, promised everything. The promise here is that a former prime minister will keep AI safe. The reality is that he will keep the companies safe from regulation. Check the block height. The transaction is already confirmed. The question is who pays the final gas fee.