The European Ombudsman has taken up the question of how the European Commission appointed Jim Hagemann Snabe — chairman of Siemens' supervisory board — to serve as an artificial intelligence envoy. According to the short report from Crypto Briefing, the inquiry concerns the transparency and propriety of that appointment. There is no published case number. No named complainant. No terms of reference I have been able to locate. Three sentences of fact wrapped around a very large question.
That thinness is the story. The thing under investigation — whether a man with a commercial stake in AI regulation helped shape AI regulation — is the oldest problem in the room where rules get written. And it is one the crypto industry has spent a decade pretending cryptography could solve.
The Ombudsman is not a court. Its mandate covers maladministration — procedural sloppiness, opacity, failure to follow an institution's own rules. It can recommend, request, and publish. It cannot compel. That distinction matters for calibration: whatever emerges from this will be a document, not a verdict. It functions as a reputational instrument — a public note that something procedural did not add up.
Snabe's résumé makes the structural tension easy to read. He chairs the supervisory board of Siemens, an industrial group that sells AI-adjacent software into European factories and therefore sits squarely inside the blast radius of the EU AI Act. He has also chaired A.P. Moller-Maersk and holds a long-standing role in the World Economic Forum's governance circuit. The AI Act is in its phased implementation window right now — prohibited practices first, then general-purpose model obligations, then high-risk system duties — each stage carrying standards that must be drafted, consulted on, and finalized.
Those standards cost money. They set the price of compliance per model, per deployment, per vertical. Siemens has opinions about them. So does every industrial AI vendor on the continent. Meanwhile Washington has moved in the opposite direction, and Brussels is openly anxious about competitiveness — hence the simplification packages, the trimming of timelines, the drift toward 'AI sovereignty' language that sounds like strategy but often means subsidy. None of that is illegal. All of it is directional.
Crypto readers will recognize the shape. We watched MiCA move through the same machinery: consultations, expert groups, technical standards, then a licensing regime. We watched who showed up to write them. The crypto press is covering this AI story for reasons larger than AI.
Europe's stated advantage in the AI race was never compute. It was rulemaking credibility. That is the asset under examination here. If the arbitration between the regulator and the regulated happens in unlisted rooms, the credibility is a marketing claim rather than a mechanism.
Start with the job itself. An envoy is typically informal, unpaid, and outside the procurement process. That structure is not incidental.
Formal appointments trigger machinery. Declarations of interest. Screening. Sometimes a public register. Informal advisory roles frequently trigger none of it, and they sit closer to the principal — an envoy briefs at the top of the institution, where agenda-setting happens before anything reaches a drafting committee. The influence arrives upstream of the paperwork.
Based on my audit work in 2017, when I spent six months pulling apart seventeen ICO whitepapers, the most dangerous vulnerabilities were never on-chain. They were in the governance section — usually blank, or filled with advisor names whose other positions went undisclosed. Three of the contracts I flagged were exploited within eighteen months. None of the disclosures that would have caught them existed, because nobody had asked.
The protocol world has not gotten much better. Watch how grant councils form. Watch how multisig keys distribute. A five-of-nine council where three signers sit inside the same ecosystem fund is not a technical flaw — the signing works fine. It is a verification gap. The chain records who signed. It records nothing about what they wanted. The uncomfortable inference from the Commission's choice is that informal roles are attractive precisely because they leave no audit trail. That is a design decision, whether or not anyone made it consciously.
This is where my current work keeps landing. At Veritas Protocol we built zero-knowledge authorship proofs for independent journalists, then ran straight into the ceiling. Proving a human typed a sentence is a solved problem. Proving that the human had no stake in the sentence is not. Code doesn't audit intent. It verifies the claims you hand it and nothing else. That gap is where every governance scandal lives.
Which brings us to why this story is being ignored. It is a bear market. Attention follows price, and a procedural inquiry into an unpaid advisory role moves no candles. Nobody is bleeding LPs over it. But the failures that kill protocols rarely arrive with a chart attached. When I helped build the Terra/Luna post-mortem on narrative decay, the finding was simple and uncomfortable: broken promises erode trust faster than broken code, and they do it quietly, in the months when nobody is watching. Bear markets do not produce fewer governance failures. They produce fewer witnesses.
'Regulatory capture' is the headline everyone reaches for. I think it is the wrong frame. Siemens did not need a backdoor into Brussels; it has a lobby and a law firm. Nobody disputes that Snabe is qualified. The dispute is that we cannot see what he is optimizing for.
The crypto-native reflex — put the interest register on-chain, prove recusal with a zero-knowledge proof — is mostly theater. A ZK proof of recusal can show you did not sign a particular document. It cannot show you did not shape it over dinner in Munich. We like to say that soulless finance is just empty pixels. Soulless governance is the same thing with a letterhead, and it costs more.
The next regulatory fight will not be about models or tokens. It will be about the legibility of the people who write the rules for both. Brussels has an ombudsman who can at least open a file. Most jurisdictions do not. If we cannot verify the alignment of one unpaid envoy, the honest question is not how we will certify a model — it is why we ever assumed we could.