Free fact-check tool

AI Hallucination Detector

Paste AI-generated text, extract claims, and verify each one against the live web. See True, False, or insufficient evidence with sources.

Hallucination Detector

Claim-level fact check

  • Claim extraction
  • Live web grounding
  • TRUE / FALSE

Developer experience

Verify AI claims with live web data.

Use the Answers endpoint to fact-check AI-generated claims by grounding them against live web sources.

REST APIAI reasoningFact verificationWeb groundingSource citationsTRUE / FALSE
Read the docs
curl --request POST \
  --url https://api.olostep.com/v1/answers \
  --header 'Authorization: Bearer YOUR_API_KEY' \
  --header 'Content-Type: application/json' \
  --data '{
  "task": "Is this claim true or false? OpenAI was founded in 2015 by Sam Altman and Elon Musk. Verify with sources."
}'

Ground AI text before you publish it

The detector splits text into claims and checks each one with the Olostep Answers API — the same true/false task as the snippet on this page.

  • Claim extraction

    A first pass pulls checkable statements out of the pasted text.

  • Web verification

    Each claim is checked against live sources via /v1/answers.

  • Suggested fixes

    False claims come back with a corrected sentence when the model can offer one.

How it works

Each tool is designed around a direct job: provide the input, get a useful output, and move it into your workflow.

  • 1

    Paste the text

    Use an AI draft, a paragraph, or a set of bullets.

  • 2

    Extract and verify

    Claims are listed, then each one is grounded on the web.

  • 3

    Review the verdicts

    Keep True claims, fix False ones, and investigate insufficient ones.

Built for practical workflows

These tools are small on purpose: they solve the conversion job quickly, then point teams toward the API path when the workflow becomes recurring or high volume.

  • Content QA

    Check a draft before it goes to a customer or a blog.

  • Research notes

    Verify stats and founding dates in a briefing.

  • Model evals

    Spot-check answers from an internal agent.

Need this at scale?

Call POST /v1/answers with a true/false task per claim. The snippet on this page is the production shape.

Frequently asked questions

Product & Capabilities

  • A claim in the pasted text that the live web does not support, or that it contradicts.

  • Treat verdicts as research signals. Ambiguous topics may return insufficient information.

  • No. It extracts individual claims and verifies them separately.

Usage & Automation

  • POST /v1/answers with a true-or-false task, matching the site’s search route.

  • Yes. Extract claims in your app, then call Answers per claim.

  • No long-term store on this page. Runs are one-off.

Pricing & Plans

  • The page UI is free to use. Live runs use the Olostep API configured for this site and may be rate limited.

  • Olostep is usage based. See pricing for current plan details and included credits.

  • New Olostep accounts include free credits so you can test extraction before scaling.

Olostep