About the Veridi Fact-Checker Bot

If you arrived here from your server logs after seeing a Veridi Fact-Checker/1.0 user-agent, this page tells you what the bot is, what it’s doing, and how to allow or block it.

What it is

Veridi is a published fact-checking methodology and a small operating service that applies it. When a user submits a claim for verification, the methodology requires retrieving primary sources, evaluating their credibility, and grounding the resulting verdict in retrievable evidence rather than the model’s training-corpus knowledge. The bot is the fetch-side of that retrieval step.

User-agent string

Veridi Fact-Checker/1.0 (+https://veridi.org/en/bot; contact via https://veridi.org/en/contact/)

What it does

  • Issues web searches against the public web through standard search APIs.
  • Fetches URLs cited or surfaced by those searches to read the underlying content.
  • Cites your URL in the fact-check output so users can visit your content directly.
  • Visits one page per cited URL per fact-check. The bot is not a crawler: it does not follow links recursively, enumerate site directories, or maintain a persistent index.

What it does not do

  • Republish your content. Pages fetched are summarized in private fact-check outputs delivered to the requesting user; the underlying text is not redistributed.
  • Train models on your content. Veridi does not collect data for model training. The published methodology runs on commercial language models that have their own (separate) training data policies.
  • Hammer your servers. Per-claim retrieval is bounded by tier-specific search budgets (typically 1–8 searches per claim, with 3-second-or-greater request spacing). The bot does not run continuous crawls.

Robots.txt compliance

The bot honours robots.txt. If you want to block it specifically, add:

User-agent: Veridi Fact-Checker
Disallow: /

Standard wildcard rules apply if you want to block all bots by default while allowing specific ones.

Why allow it

Fact-checking is a public-interest function. When users submit claims about your content (for example, claims that cite you as a source, or claims that reference reporting you’ve published), Veridi needs to retrieve the actual page to compare the claim against your content rather than rely on a language model’s recollection. The fact-check output includes the cited URLs, so users can read your content for themselves. Blocking the bot means those fact-checks will be processed without reading your content directly. The result is typically an INSUFFICIENT EVIDENCE or UNVERIFIABLE verdict with a LIMITATIONS note recording the block. The methodology does not substitute fabricated retrieval.

Why we identify ourselves

We don’t use a browser user-agent to bypass anti-bot defenses. The user-agent above is the only one we use, including when it’s blocked.

Contact

Methodology questions, abuse reports, allowlist requests, or operator feedback: please use the contact form.

Version history

  • 1.0 (2026-05-02): initial declared user-agent. Replaces the prior Veridi Fact-Checker/1.0 string that had no public-info URL.