Do I need an independent bias audit for my AI interview tool?

If you hire in New York City, yes. NYC Local Law 144 requires an independent bias audit completed within the prior 12 months, a public summary of the results on your site, and at least 10 business days' notice to candidates before you use an automated employment decision tool. The audit must be run by an independent auditor, not the vendor.[1]

The EEOC withdrew its AI guidance, am I off the hook?

No. In early 2025 the EEOC removed two AI technical-assistance documents from its site, but Title VII, the ADA, the ADEA, and the four-fifths rule remain fully binding. Only the non-binding guidance was withdrawn; the underlying law is unchanged and there is no safe harbor. Private plaintiffs can still bring claims.[2][6]

Am I liable for a vendor's tool?

Usually the employer is liable, and the vendor can also be liable as an agent (a theory advancing in Mobley v. Workday). That is why you should collect evidence from your vendor: audit methodology, selection-rate data, and logging guarantees.[2]

When does the Colorado AI Act take effect?

The original SB 24-205 never took effect; it was replaced by SB 26-189 with a new effective date of January 1, 2027. The new version drops the duty-of-care and impact-assessment mandates for a narrower notice-and-transparency model, enforced by the Attorney General with no private right of action.[5]

Does Intrvio have a published bias audit?

Intrvio produces audit-ready evidence: selection-rate data, transcript-anchored scores, candidate notices, and human-review records. The independent audit is run by an auditor you choose (for example Holistic AI, BABL AI, DCI Consulting) for your deployment, and you publish the result; Intrvio one-click exports the data that audit needs.[1]

US AI hiring compliance: NYC LL144, EEOC, Illinois, Colorado

Is AI hiring legal in the US?

AI hiring is legal in the US, but it must comply with federal anti-discrimination law. Title VII (race, color, national origin, religion, sex), the ADA (disability), and the ADEA (age 40+) apply to AI tools, and the disparate-impact theory holds. The four-fifths (80%) rule in 29 CFR 1607.4(D) is the baseline measure for selection-rate disparities.[2]

An important nuance: the EEOC removed two AI technical-assistance documents in early 2025, but the laws themselves did not change. Only the non-binding guidance was withdrawn; there is no safe harbor and private plaintiffs can still sue. Vendors can also be liable as agents (Mobley v. Workday).

NYC Local Law 144

NYC Local Law 144 requires three things before you use an automated employment decision tool: an independent bias audit completed within the prior 12 months, a public summary of the results on your hiring page, and at least 10 business days' notice to candidates. The published summary must include the audit date, the data source, the number of applicants, the selection/scoring rates, and the impact ratios for sex, race/ethnicity, and intersectional categories.[1]

The penalties look small but stack: $500 for a first violation, $500 to $1,500 for later ones, and each day of non-compliant use plus each failure to notify counts as a separate violation. An independent audit runs roughly $5,000 to $30,000 per tool in the market; recognized auditors include Holistic AI, BABL AI, DCI Consulting, ORCAA, and BLDS.[1]

State laws to track in 2026

Illinois. The AI Video Interview Act (820 ILCS 42) requires notice, explanation, and consent.[4] Public Act 103-0804 (Jan 1, 2026) makes AI-based discrimination, using ZIP code as a proxy, and failing to disclose AI use a civil-rights violation.[3]
Colorado. The original SB 24-205 never took effect; it was replaced by SB 26-189, effective January 1, 2027. Enforced by the Attorney General with no private right of action; up to $20,000 per violation under the state consumer-protection act.[5]
Other states. Maryland requires written consent for facial recognition; Connecticut PA 26-15 phases in over October 2026 and 2027; California ADS employment regulations have been in force since October 1, 2025; Texas TRAIGA took effect January 1, 2026.[7]

Employer obligations checklist

If you hire in NYC, get an independent bias audit before use and publish the summary.
Notify candidates that AI is used (10 business days ahead in NYC, with consent in Illinois).
Assign a human-oversight step where an authorized person makes the final decision.
Keep selection-rate data and decision records ready for four-fifths analysis.
Collect methodology, logging, and change-notification evidence from your vendor.

How Intrvio helps

Intrvio treats compliance as a by-product of daily operations, not a badge you buy. To be clear, the independent audit is run by an auditor you choose and you publish the result; Intrvio produces and one-click exports the evidence that audit needs.

Audit-ready evidence. Selection rates, transcript-anchored scores, and decision records are exportable.
Candidate notice. A notice that AI is used, with a human-review path template.
Human oversight. GAIA produces evidence; a human makes the decision, recorded separately.
Region tagging. Every record carries a US region tag, proving data residency in audits.

Responsibility is shared

Let us be plain: Intrvio is a tool, the employer is the decision-maker, and this page is not legal advice. Work with your legal team. For the EU see EU AI Act compliance, for Turkey see KVKK compliant interviews. To see the product, visit AI interview platform.

US AI hiring compliance

Is AI hiring software legal in the United States?