Auditing of algorithms can help address bias but there needs to be industry standards/regulations to hold the auditors or the companies that use them to account
Context & Ripple Effects
Algorithmic auditing has been pitched as the accountability mechanism of record since Democratic senators proposed the Algorithmic Accountability Act in 2019, which would have required large companies to audit their machine-learning systems for bias. Skeptics countered in early 2021 that audits alone won't deliver accountability because they lack market incentives and real government oversight.
What changed is that mandates started creating real demand: New York City now bars employers from using AI hiring tools unless a yearly bias audit clears them. The Markup's argument lands on the gap that creates — an audit requirement without standards for who audits, how, or what happens when findings are ignored.
First-order effects
- New York City employers deploying AI hiring tools must now commission annual bias audits, making auditors de facto gatekeepers over whether those tools stay in use at all.
- Companies buying audits face wildly uneven quality: with no accreditation or methodology standard, the same system can pass one auditor's review and fail another's.
Second-order effects
- An audit-for-hire market grows around every new mandate like NYC's, but because auditors are paid by the companies they assess, the incentive structure mirrors the conflict-of-interest problems regulators spent decades policing in financial auditing.
- Governments procuring third-party algorithms — already flagged by researchers as overpaid and under-supervised in fraud-detection contracts — get no protection from a biased vendor's clean audit report unless the audit itself is standardized.
Third-order effects
- If the pattern holds, regulation evolves from mandating audits to regulating auditors — accreditation regimes, liability rules, and disclosure standards that turn ad hoc bias reviews into a licensed assurance profession.
- The alternative outcome, warned by both the Fast Company critique and Wired's call for external oversight, is a compliance theater equilibrium: audits exist on paper while biased systems ship anyway, hardening the case for direct regulatory enforcement rather than self-reported review.
The trend: AI accountability is shifting from voluntary, company-commissioned audits toward mandated third-party assurance — with the next battleground being who accredits the auditors.