Methodology
This page documents how AI Trace operates: our editorial process, evidence standards, data model, and commitments.
Editorial Process
Entries go through a four-stage lifecycle: submission, review, publication, and ongoing verification.
Submission
Anyone can submit a report via the Report a Sighting form. Reports include a description of what was observed and an optional evidence link or screenshot. No account required.
Review
A moderator reviews each submission. They verify the claim against the provided source, assess whether the source meets our evidence standards, and check for duplicates.
Publication
Accepted submissions become structured entries linked to the company profile. Each practice gets a category, severity tag, and at least one verified source.
Ongoing Verification
Entries are periodically re-checked. If a company reverses a practice or new contradicting information emerges, entries are updated with full edit history preserved.
Every change to every entry is recorded in the public edit history visible on each company profile page. This append-only log is the primary mechanism against astroturfing.
Evidence Standards
Every claim must be backed by at least one cited source. We categorize sources into six types, ranked by reliability:
Company Disclosure
Official statements, blog posts, press releases, or SEC filings from the company itself. Highest reliability.
Regulatory Filing
Documents filed with government agencies (FTC, EU regulators, NLRB). Carries legal weight.
Academic Paper
Peer-reviewed research documenting AI practices. High reliability but may lag real-world deployment.
News Article
Reporting from established journalists. Accepted if the publication is reputable and uses primary sources.
Social Media Post
Posts from company accounts or verified employees. Lower confidence. Screenshots required.
Community Report
User-submitted observations. Always marked as unverified until corroborated by another source type.
We reject claims based solely on speculation, rumors, or sources that cannot be independently verified. When evidence is disputed, the disputed status is shown on the entry.
Neutrality Commitment
We do not tell you how to feel about AI. We document what companies do and let you draw your own conclusions.
The UI does not use red/green good/bad framing. Status badges are factual, not emotional. “Confirmed AI Use” is descriptive, not a judgment. We track AI use; we do not rate it.
We do not accept payments from companies to influence their entries. Companies cannot submit corrections directly; they must go through the same community process as everyone else. All edits are logged and public.
Severity tags describe real-world effects, not moral judgments. A practice tagged “Replaces Human Labor” is factual, not an accusation.
Data Model
Every AI practice entry is structured with the following attributes:
Dispute Process
If you believe an entry is inaccurate, you can file a dispute directly from the practice detail page. Disputes are reviewed by moderators and resolved transparently.
Dispute types include: wrong or fabricated information, outdated practice, missing context, duplicate entry, and source reliability concerns. Every dispute resolution is logged in the public edit history.
We do not silently remove contested information. If an entry is disputed, it carries a “Disputed” badge until the dispute is resolved.
How to Contribute
The most valuable thing you can do is submit evidence. If you notice a company using AI, report it via the Report a Sighting form. It takes under 90 seconds. No account required.
You can also help by sharing news articles about corporate AI practices, linking to company profiles in relevant discussions, and filing disputes when you spot inaccuracies.
Our Use of AI
AI Trace tracks corporate AI use. We hold ourselves to the same standard. Our own AI practices are documented on our company profile at /company/trace-foundation.
Intelligent Submission Processing
ActiveWhen a community submission arrives, Claude API (Haiku) suggests which company it refers to (entity matching), whether it duplicates an existing practice, and what category it should be. These suggestions appear as pre-filled, editable fields in the moderator review interface. Moderators always have the final say.
Semantic Search
ActiveVector embeddings (OpenAI text-embedding-3-small) power semantic search alongside keyword matching. Users searching for concepts like "AI voice acting" find relevant results even when entries use different terminology. The search experience is invisible to users; results are simply more relevant.
Automated Evidence Discovery
PlannedA future feature that will automatically scan public sources for new evidence of corporate AI use and surface potential entries for moderator review. This feature is in the planning stage and will not be built until grant funding is secured. When active, all discovered entries will still require human moderator approval before publication.
Nothing AI-assisted is published without human moderator approval. All AI suggestions are clearly labeled and can be overridden or ignored by moderators.
For information about the team and how to support our work, see the About page.