FairAudit
Menu

Audit your hiring AI for fairness.

Independent bias audits, evidence traced to the EU AI Act and NYC Local Law 144, and compliance checks your engineers see where they work.

Apply for early access Try the free EU AI Act assistant

Regulatory information, not legal advice.

Meritocracy doesn’t happen by declaring it. It happens by measuring it.

Companies that announce they reward merit can end up more biased, not less (Castilla & Benard, 2010). Hiring AI can be fairer than people, but only when someone checks. We measure it, so no one is left out by a model nobody looked at.

Sources →

FairAudit Audit Early access

An independent bias audit of your hiring AI.

We measure selection rates and impact ratios for every group the law names, using the method NYC’s bias-audit rule prescribes, on your real outcomes. You share counts per group, never personal records. You get a signed report your clients’ DPOs can verify.

Apply for an audit How an audit works →
Sample report, synthetic data Decision point: Shortlist · engine fairness/1.1.0
Gender 4,050 candidates
Group Selection rate Impact ratio
man reference 19.6% 1.00
woman 16.2% 0.83
non-binary 14.0% 0.71
Impact ratio: a group's selection rate divided by the best-treated group's. 0.80 is a benchmark, not a verdict. A named person makes the determination.
Sample report, synthetic data. Counts invented for this page. Computed by the FairAudit fairness engine. Not any client and not a published audit.

FairAudit Flow Early access

Compliance where your engineers write tickets.

While a ticket is written, Flow suggests acceptance criteria tied to the article they come from, asks at most three questions about facts only your team knows, and logs every decision as evidence.

Join the Flow pilot How Flow works →
Two suggestions from Flow. The second is tied to Article 50(2) and Article 50(1), with Apply and Dismiss buttons.
  1. 1 Draft: it arrived while the ticket was being written
  2. 2 The articles the suggestion comes from
  3. 3 Apply or dismiss. Both are logged.
Flow’s suggestions on a ticket, from a local run with no model connected.

FairAudit Assistant Free · live

Ask the EU AI Act. Get the article.

Every answer cites the exact provision. Sign in with your email. We confirm access by email.

Try it free What it answers →
The assistant answering whether AI may infer candidates’ emotions from video interviews: no, citing Article 5(1)(f), Article 3(39) and Annex III point 4, with the retrieved passages beside the answer.
  1. 1 A straight answer first
  2. 2 Each reason cites its article
  3. 3 The passages it retrieved
A real answer from the assistant, to a question from its test set.

How we stay right

Every output is signed.

Verify any FairAudit report yourself, no account needed.

Verify one →

The law is watched.

RegWatch reads official EU sources and drafts updates. A person approves every change.

How RegWatch works →

Measured, not assumed.

We score how many of the claims in our answers are supported by the passages we cite (independent judge, repeated runs). The figure for the current index is not published yet.

What it does and does not do →

Who it’s for

HR-tech vendors whose clients ask “how do I know your AI is fair?”, the employers who use them, and the AI teams who carry governance.

See where you fit →

AI shouldn’t be at the helm of HR. People should, with evidence.

The EU AI Act requires human oversight of high-risk hiring systems (Art. 14). We help you show it is real.

Apply for early access Try the free EU AI Act assistant

The assistant: sign in with your email. We confirm access by email. The high-risk obligations for hiring systems, Article 14 included, apply from 2 December 2027. Regulatory information, not legal advice.