LO
DeploymentProcess & self-service2026-01-28
The US commodities regulator reported in the federal AI inventory that it runs an isolation-forest model daily to flag potentially erroneous data loads
Data engineeroccupation page →
Event date / reported
2026-01-28
Evidence stage
DeploymentAn employer has put it into production. Can move the baseline — weighted by scale and how similar the setting is.
Tasks this bears on
When the data is wrong and nothing errored
A field changed meaning upstream, a job ran twice, a timezone shifted — and every dashboard is still green.
Still human-led✓ Evidence-backed
Where this applies
One regulator, one dataset, reported under a statutory duty. Entry CFTC-001 in the individually-reported CSV, whose whole text is two sentences: an anomaly detection model "designed to identify potentially erroneous data loads in TCR data", using "an isolation forest model with aggregated data", which "automatically runs daily". The entry carries no operational date, no vendor, no volume and no account of what the model catches or misses, so the date here is the publication of the consolidated inventory (2026-01-28) and nothing about the deployment itself is dated. It establishes that at least one data team has handed the first half of this task, noticing that a load looks wrong, to a scheduled model; it says nothing about the second half, working out what the wrong load means and what to do. The same inventory carries an entry that reads the other way: DOJ-0006, the Bureau of Alcohol, Tobacco, Firearms and Explosives, lists a data platform whose anomaly detection and data-quality features are "embedded within the tools" and adds "while we do not use these features directly". A listed capability and a used one are different things, and the inventory format lets an agency say which it has.
What this means
The first half of this task, noticing that a load looks wrong, is precisely what an isolation forest run daily over loads does, and one regulator runs one. What that leaves is the half a model cannot do: working out what the anomaly means, whether it matters, and what to change upstream. That half is the one that keeps the dashboards honest, and nothing in this entry touches it.
What it does not yet show
Two sentences, no date, no volume, no precision. Nothing about what the model catches or misses, whether anyone acts on its flags, or how often it is wrong. And the same inventory has another agency listing the same capability and saying in the next sentence that it does not use it, so a listed detector is not a used one, and this record cannot tell you which kind yours would be.
What you can check
Ask how many data-quality incidents in the last quarter were first noticed by a person and how many by a check. If nobody can answer, that is the finding: you cannot know what a model would take over until you know what people currently catch.
Does it change the assessment?
No. The impact index is never moved by a single event. What this record did: the 1 linked task judgement above now rest on evidence instead of inference.
Source
OMB — 2025 Federal Agency AI Use Case Inventory (entry CFTC-001) · verified 2026-09-20 · Claude (agent) · interpreted 2026-09-20 · Claude (agent)
Primary source — published by the party that did this, or the authority of record. No co-signature needed.