Built so a courtroom can check its work.
Who makes this
Child Evidence is built and maintained by Matthew DiBenedetto, a law student, as an independent public-interest project. Every published claim — its canonical phrasing, both evidence ratings, its guardrails, and its publication — is signed off by a human editor. The full editorial method, including exactly where AI assists and where it is excluded, is documented on the Method page.
Why it exists
Developmental science that bears on how children are treated in legal proceedings is scattered across paywalled journals and written for scientists. The attorneys who need it — often appointed, often time-pressed — need to know one thing a science search engine will not tell them: how strong is the evidence behind this claim, and will it survive an admissibility challenge? Child Evidence exists to answer that question transparently and to remain free to use.
Funding and independence
Child Evidence is unfunded and independent. It accepts no money from litigants, law firms, expert-witness services, publishers, or advocacy organizations; it sells no data and runs no advertising. If that ever changes, the disclosure will change here first and be recorded in the public ledger.
How to cite Child Evidence
For the site as a whole, in Bluebook form (Rule 18.2.2):
Matthew DiBenedetto, Child Evidence (last visited Sep 23, 2026), https://www.childevidence.com.
When citing a specific claim, use that claim’s URL and the “Last reviewed” date shown on its page — and cite the underlying studies directly wherever possible; every claim page gives you their full citations. Nothing here is a substitute for reading the study you cite.
Review process
Every published claim passes the same protocol: AI may draft, but a human editor performs each act of judgment — the canonical phrasing, both ratings (with stored criteria and rationale), the guardrails, and publication sign-off. Grounding screens run fail-closed and are advisory only; nothing publishes on a model’s say-so, and no citation can render unless it is a database reference to a verified study record. Changes of any kind are logged to the public ledger. External advisory review (a developmental scientist and a practicing attorney) is planned; when reviewers are named, they will be named here.
Corrections
Errors are fixed publicly. If a claim, grade, or citation is wrong, email corrections@childevidence.com — corrections are recorded in the ledger.