What this is
About Check AI Claims
Check AI Claims reviews public AI product claims for evidence burden and wording risk. Paste a product page or public claim text. Get an evidence-burden note, evidence gaps, buyer questions, and wording boundaries.
The tool
AI product pages often make claims about accuracy, automation, compliance, ROI, or uniqueness without explaining the evidence behind them. Check AI Claims extracts those claims from a public page or pasted text, classifies each one by type and evidence burden, and returns a structured Claim Receipt.
A Claim Receipt includes: an evidence-burden note, the evidence that would support the claim, buyer questions to ask the vendor, and wording boundaries to compare against. It is designed to help the reader ask better questions — not to render a verdict.
Who uses it
The tool is primarily for buyer-side review:
- Buyers and procurement teams evaluating AI products — checking whether vendor claims are specific enough to rely on, and generating questions to ask before purchasing.
- Due-diligence and content reviewers checking public AI claims — identifying the evidence a claim would need before it is used in a report, vendor review, or internal note.
Marketers may still use the checker to inspect public copy, but the product is not a marketing copywriting workflow.
What it is not
- a truth or fake AI judge
- a company blacklist or ranking
- a fraud or scam detector
- legal advice or a legal opinion
- a compliance certification or audit
- an investment or procurement recommendation
- a public accusation or reporting platform
What it currently supports
The checker accepts two input types: a public product page URL, or pasted marketing text up to 5,000 characters. It extracts AI-related claims from the content and returns a Claim Receipt.
It does not support PDF uploads, screenshots, private pages, authenticated dashboards, batch checking, or whole-site scanning. If a page cannot be read or contains no clear AI-related claim, the tool says so rather than forcing a result.
Cases and guides on the site draw only from official enforcement actions, regulator reports, and high-confidence published sources. User submissions are private by default and go through manual review before anything is published.
Editorial and review
Created and maintained by Shane Wei. For questions or corrections, email hello@checkaiclaims.com.
The signal catalog, source cards, reviewed cases, and methodology are maintained by the Check AI Claims editorial team. Every source card references an official source — enforcement action, regulatory guidance, research report, or published standard — with a direct link. Signal-to-source mappings, section references, and downgrade evidence are reviewed manually before being added to the catalog.
Sources are reviewed for active or superseded status. When an official page is removed or archived, the source card is updated to Superseded status with a link to the current primary law text. The signal mapping version and date are recorded on every Claim Receipt.
If you find an error in a source reference, signal mapping, or case summary, use the Corrections page to submit a source update. For general questions, use the Contact page.