Full disclosure of which AI tools AI Leakage uses, how we use them, and the limits we apply.

The short version

AI Leakage uses AI as a research and drafting tool. The primary tool is Claude (Anthropic, commercial tier with training contractually excluded). Every claim in every vendor profile is reviewed by a human before publication. AI-assisted drafts are never published without human verification of every cited source.

Primary tool: Claude (Anthropic)

We use Claude for: research synthesis (reading large numbers of sources and identifying the relevant material), fact-checking against published sources, drafting assistance, and editing for clarity. We use the commercial tier (currently Claude Pro through the API integration with appropriate no-training contract terms), not the consumer free tier.

The Anthropic vendor profile on this site applies a deliberate counter-correction to avoid favourable bias. Where the evidence on Anthropic is unflattering, we report it with the same specificity we apply to OpenAI, Google, and Microsoft. The Claude profile is one of the most-detailed in our database for this reason.

Other AI tools we use occasionally

• ChatGPT (OpenAI), commercial tier: Used occasionally for cross-checking Claude’s analysis on specific factual questions. Same no-training contractual default.
• Perplexity Enterprise: Used occasionally for research grounding on recent events where web search is necessary. Enterprise tier only, not consumer tier (the consumer-tier risk picture is documented in the Perplexity vendor profile).
• No other AI tools are used in our editorial workflow as of the date of this page.

What we never use AI for

• Final fact verification. Every cited URL is visited by a human on the verification date shown at the top of each profile. AI is not trusted to confirm whether a URL exists or what its content is.
• Risk ratings. The 1-to-5 risk rating for each vendor is set by a human after reviewing the evidence. AI may help structure the reasoning; it does not assign the rating.
• Editorial recommendations. The “Recommended for” and “What this means in plain English” sections of each profile are written and reviewed by a human. AI may help with phrasing; the substance is human judgement.

If you have concerns about our AI use

Reasonable people can disagree about whether a site critiquing AI vendors should use AI tools in its workflow. Our position: refusing to use AI for research would not produce better content, and pretending we do not use it would be dishonest. The right response is transparent disclosure, demonstrable human editorial control, and a profile of our primary AI vendor that is more skeptical than it would otherwise be.

If you spot a claim on this site that appears to be an AI hallucination — a fabricated CVE number, an invented incident, a misquoted source — please email the operator via the contact page. We will verify, correct, and publish a correction note.