Non-Targeting
Nothing here is directed at any identifiable person or organisation.
This site argues for a legal mechanism that does not exist. Everything below follows from that, and I want it stated without hedging, because the subject matter makes the hedged version worse than useless.
There is no bounty
No bounty described on this site is offered, funded, or collectable. There is no fund, no claims process, and no party who would pay. The penalties I discuss are hypothetical statutory penalties in a hypothetical statute. Nobody can violate a law that has not been written, so nobody described here has done anything wrong under it, and no reward is available to anyone for anything.
Nobody is being asked to investigate anyone
The mechanism I describe would, if enacted, pay people for evidence. That is a claim about how a future legal regime could be designed. It is not a request, and it is emphatically not a request to start now. I am not asking readers to surveil colleagues, collect documents, infiltrate laboratories, approach insurers, or contact authorities about any person or organisation. Private policing in the absence of a legal framework is exactly the failure mode the policy-research disclaimer on every page warns about, and I would consider anyone acting on this material that way to have badly misread it.
Named parties are named for public reasons
Where this site names a company, a laboratory, a researcher, or an incident, it does so on the basis of public reporting, published research, or the party’s own disclosures, and cites the source. Those references illustrate a general argument about incentives. None of them is an allegation that the named party has broken any law, breached any duty, or acted in bad faith. Where I describe conduct I think is risky, I mean risky in the ordinary sense that reasonable people disagree about frontier research — not culpable.
Several passages describe hypothetical people: an unnamed researcher at an unnamed lab, a firm that does not exist, a scenario constructed to make a mechanism legible. These are constructed examples. Resemblance to any real person or organisation is not intended, and where a real case would fit the shape of an example, that is a fact about the shape of the example.
If you think this is about you
It is not. If something here nonetheless reads as though it describes you or your organisation, and you would like it changed, clarified, or removed, please get in touch . I would rather correct a page than leave someone feeling accused by it. That offer is open regardless of whether I think the reading is fair.
What I am actually claiming
That the current incentive structure around frontier research is badly shaped, that mechanism design is a more promising lever than exhortation, and that this particular mechanism deserves more attention than it has had. Those are claims about institutions. They are not claims about the character or conduct of the people working inside them, most of whom I take to be acting in good faith on a problem they did not create.