Disclosure Obligation
A legal requirement to publish a figure, which turns out to predict that figure's reliability better than how much the answer matters.
Reviewed July 16, 2026Stable
When not to use it
- To dismiss unregulated figures. Most are produced carefully; they are unverifiable rather than untrue.
- Where a mandate exists but its definitions no longer match the question, in which case compliance and usefulness diverge.
- As an argument that more regulation always improves evidence, since a badly specified mandate produces comparable numbers about the wrong thing.
Reach for something else instead
- Voluntary structured reporting — a stated method with published limitations, which is often the best available and is not enforceable.
- Independent replication — the substitute for obligation where none exists, and the thing that almost never happens at scale.
- -
Read more on the blog
- The EU delayed the part with no standardsThree AI Act obligations take effect today and the widely reported headline says the opposite. What was deferred, what was not, and why the split falls exactly where it does.
- The framework that exists produced 1.6%The FDA has two AI tracks. One is final, has authorised over 1,350 devices, and is the regime under which almost none of them cite a trial. The other missed its own deadline five weeks ago.
- Zero-click is 60%, or 22.4%, from one providerTerritory 9 opens on what cheap generation does to information. Every method agrees the traffic is falling. None agrees on how far, and the headline metric differs threefold within one dataset.
- The 12% was non-inferior, and P was 0.41MASAI is the best-evidenced AI deployment in medicine and the headline everyone quoted describes a result the trial did not claim.
Further reading
- International Energy Agency (2025), Energy and AI — a case where a research body measured what no regime required.
- Wong et al. (2021), External Validation of a Widely Implemented Proprietary Sepsis Prediction Model — deployment at scale with no obligation to validate, and what an independent check found.
Primary sources, listed so you can check the claims on this page rather than take them on trust.
Where people go wrong
- Treating a figure's precision as evidence of its verifiability.
- Comparing numbers from different tiers as though they were equivalent.
- Assuming an important question is well measured because it is important.
At a glance
FieldSafety & Ethics
Predictsreliability, better than importance does
Three tiersmandated, voluntary structured, unstructured
Secondary benefitcomparability through fixed definitions
Failure modereliable answers to an obsolete question
DifficultyBeginner → Intermediate
Where this sits
A starting point. Nothing needs to come before it.
0Levelstarting point
0Needs firstnothing
0Opens upnothing further
1Areastays here
Computed from the prerequisite graph, not assigned. How this works