AI Understanding Data API

Quickstart

No signup for the free tier. Every response carries a license field and an X-Attribution header with the exact credit line to reproduce.

Free tier — no key

curl -s "https://aiunderstanding.org/api/v1/news?limit=5"

Keyed request

curl -s "https://aiunderstanding.org/api/v1/funding?limit=100" \
  -H "Authorization: Bearer aiu_live_YOUR_KEY"

Cursor pagination

# Page through full history (keyed plans)
curl -s "https://aiunderstanding.org/api/v1/news?limit=100&cursor=NEXT_CURSOR" \
  -H "Authorization: Bearer aiu_live_YOUR_KEY"

Endpoints

All endpoints are GET, return JSON, allow any origin, and accept the key as Authorization: Bearer …, X-API-Key or ?api_key=.

GET /api/v1/news

Every published story with its canonical URL, category, publication and update timestamps, cited source, hero image URL and byline.

category — One of Breaking, Product, Innovation, Enterprise, Media, Policy, Security, Industry.
limit — Page size. Free: up to 20. Pro: 200. Enterprise: 1000.
cursor — Opaque cursor from nextCursor. Keyed plans only.

GET /api/v1/releases

Release claims parsed from our coverage: mentioned organizations, selection evidence, source evidence links and suggested follow-up dates. Availability is never assumed to be verified.

organization — Watchlist id, e.g. openai, anthropic, google, meta, mistral.
limit — Page size.
cursor — Opaque cursor from nextCursor.

GET /api/v1/funding

Funding rounds parsed from headlines: company, normalized amount, currency, round label, best-effort investors, and the article the row came from.

currency — Three-letter code, e.g. USD, EUR, GBP.
limit — Page size.
cursor — Opaque cursor from nextCursor.

GET /api/v1/regulation

Laws, executive orders, guidance, enforcement and court actions from our Policy and Security coverage, with jurisdiction, instrument and headline status keywords.

jurisdiction — EU, US, UK, China or other.
limit — Page size.
cursor — Opaque cursor from nextCursor.

GET /api/v1/glossary

The full plain-language AI glossary: term, definition and the canonical term page URL.

q — Case-insensitive substring match on term and definition.
limit — Page size.
cursor — Opaque cursor from nextCursor.

Response shape

Every dataset answers with the same envelope, so one parser handles all five.

{
  "dataset": "news",
  "generatedAt": "2026-09-03T12:00:00.000Z",
  "license": {
    "name": "AI Understanding Data License 1.0",
    "attributionRequired": true,
    "attribution": "Data from AI Understanding (https://aiunderstanding.org)",
    "linkBack": "https://aiunderstanding.org",
    "terms": "https://aiunderstanding.org/developers#terms"
  },
  "plan": { "id": "free", "requestsPerHour": 60, "fullHistory": false },
  "count": 5,
  "totalAvailable": 20,
  "nextCursor": null,
  "items": [
    {
      "id": "example-story-slug",
      "title": "…",
      "excerpt": "…",
      "category": "Policy",
      "publishedAt": "2026-09-02T09:14:00.000Z",
      "url": "https://aiunderstanding.org/news/example-story-slug",
      "sourceUrl": "https://source.example.com/article",
      "sourceName": "Source Name",
      "imageUrl": "https://aiunderstanding.org/api/news/image/example-story-slug?v=…"
    }
  ]
}

Errors

# No key, more than 60 requests in an hour
HTTP/1.1 429
{"error":{"code":"rate_limited","message":"Free tier is limited to 60 requests per hour per IP…","retryAfterSeconds":812}}

# Unknown or revoked key
HTTP/1.1 401
{"error":{"code":"invalid_api_key","message":"That API key is not recognised…"}}

# Monthly quota used up
HTTP/1.1 429
{"error":{"code":"quota_exceeded","message":"Monthly quota of 100000 requests is used up…","retryAfterSeconds":94211}}

Free-tier terms

Attribution is required. Credit “Data from AI Understanding (https://aiunderstanding.org)” wherever the data appears, including a clickable link back to aiunderstanding.org.
Rate limit: 60 requests per hour per IP address, and the latest 20 items per dataset. Cursor pagination into full history needs a key.
No republishing wholesale. Use the data in your own product, research or analysis. Do not mirror the archive as a substitute for the site.
Records describe our coverage. Records describe what our newsroom published. They are not independently re-verified by this API and carry no warranty of completeness.
Editorial independence. Paying for a key buys request volume and nothing else. It does not influence what we cover or how. See our editorial standards and earnings disclosure.
Fair use of the free tier. Sustained scraping past the limit gets throttled, not billed. If you need more, get a key.

Plans

Paid plans are invoiced monthly. Revenue supports free AI education and the newsroom.

Free

$0 no key required

60 requests per hour per IP
Latest 20 items per dataset
Attribution and a visible link back are required
All five endpoints, public JSON, CORS open

Pro

$99 per month

100,000 requests per month
Full archive history with cursor pagination
Up to 200 items per page
Attribution still required in published work

Enterprise

$999 per month

2,000,000 requests per month
Up to 1000 items per page
Bulk backfill and custom fields on request
Named contact for schema changes

Request a key

Tell us what you are building. We create keys by hand, so include enough detail for a person to say yes.

The same datasets are published as human-readable research pages with full methodology at /research. Sponsorship and newsletter placements are on /sponsorship. Questions: [email protected].

Our newsroom's data, as plain JSON.