How AI agents monitor DOJ enforcement actions
An AI agent monitors DOJ enforcement by calling a wire that reads the Justice Department news feed at runtime. It returns indictments, settlements, guilty pleas, sentencings and antitrust actions with a publication timestamp and a link to the announcement, filtered by keyword and time window.
/mcp. Nothing here needs an API key or a signup.
DOJ output is high volume and mostly local. The feed carries announcements from the Office of Public Affairs alongside the 93 US Attorney offices, so an unfiltered query returns a district-level fraud prosecution far more often than anything with national reach. Volume is why this source rewards a keyword query and punishes a category subscription.
Query by party, not by category
There is one feed behind this wire, so the src filter takes a single value and does no useful work. q is the filter that matters, and the terms that pay off are entity names rather than crime types.
- Company and person names catch the announcements that concern you regardless
of which district issued them.
- Program or statute names such as a False Claims Act term or a sanctions
program catch a category of action across districts.
- Broad crime words such as "fraud" return dozens of unrelated local
prosecutions a week and are the fastest way to build a monitor nobody reads.
One quirk to code around: the feed mixes press releases with calendar entries from the Office of Information Policy, and those event items carry future publication dates. An agent that sorts strictly by timestamp will surface a workshop scheduled for October above an indictment filed yesterday. Filter on the item link path or drop items dated in the future.
The call an agent makes
# free preview, top 3 results, no payment and no wallet curl "https://thebotwire.com/doj/preview?q=antitrust&since=7d" # paid, $0.005: everything DOJ announced about a company in 14 days curl "https://thebotwire.com/doj/latest?q=acme&since=14d&limit=20" # a statute or program, across districts curl "https://thebotwire.com/doj/latest?q=false%20claims%20act&since=14d"
Parameters are q (company, person, statute or program), since (30m, 2h, 24h, 7d, 14d) and limit (1 to 50). The wire keeps a 14-day retention window and defaults to a 7-day lookback. The route costs $0.005 per call in USDC on Base, with no API key and no signup, so a daily sweep over ten watch terms runs about $1.50 a month.
An indictment is not a conviction
This is the sentence that separates a compliance agent from a defamation risk.
| Stage | What DOJ announced | What is established |
|---|---|---|
| Indictment or complaint | Charges filed | An allegation. Nothing is proven |
| Guilty plea | Defendant admitted the conduct | Established as to that defendant |
| Conviction | A verdict returned | Established at trial |
| Civil settlement | Parties resolved a claim | Usually no admission of liability |
DOJ press releases carry a standard line noting that a charge is an allegation and the defendant is presumed innocent. That line is in the document and it is not in the feed metadata, so an agent that summarises only the headline strips out the one qualifier the announcement was careful to include.
The volumes make the stakes concrete. False Claims Act settlements and judgments exceeded $6.8 billion in fiscal year 2025, and whistleblowers filed 1,297 qui tam suits, the highest count on record. Most of those suits produce no finding of liability. An agent that reports a filed complaint as a proven fraud is wrong far more often than it is right.
The safe output is the same shape used across the government wires: name the action, name the stage, give the date, link the announcement. Related signals arrive on the court opinions wire when a matter is decided and on the federal regulations wire when a rule follows an enforcement trend. Which endpoint owns which question is listed in the routing table.
FAQ
Can an agent get every DOJ enforcement action from this wire?
It gets what the Justice Department publishes to its news feed, which is the great majority of what the department announces publicly. It does not get sealed matters, it does not get filings that generate no press release, and it does not get court dockets. An empty result means nothing was announced in that window, not that nothing happened.
How does an agent avoid drowning in local prosecutions?
Filter with q on entity names rather than crime types, and keep the watch list short. A ten-term list run once a day is ten calls and stays readable. A subscription to the whole feed is hundreds of items a week, almost none of which concern the entities you track.
Does this cover the SEC or the FTC?
No. This wire carries justice.gov announcements. SEC and FTC enforcement sit on a separate enforcement wire, and financial disclosure filings sit on the EDGAR wire covered in can ChatGPT read live SEC filings. Stating the boundary is better than letting an agent infer coverage it does not have.
How fresh is the data?
This wire sits in the sparse ingest tier, polled at most every 15 minutes, which is well inside the department's publishing cadence. The observed interval between ingest runs on this service is about 294 seconds and the last poll time is published live at /health.
Sources
Enforcement volumes on this page come from the Justice Department press release False Claims Act Settlements and Judgments Exceed $6.8B in Fiscal Year 2025, which reports settlements and judgments exceeding $6.8 billion for the fiscal year ending September 30, 2025, and 1,297 qui tam suits filed, breaking the prior record of 980 set in 2024.
Related: How do AI agents read new federal court opinions? · How do AI agents track new US federal regulations?