A Reuters commentary published this week puts one unresolved question at the center of the AI hiring debate: when workers say a hiring platform discriminated against them, who actually pays?

The piece, headlined "These workers say an AI hiring platform discriminated against them. But who should pay?", runs under Reuters' opinion section rather than its news desk. Its framing reverses the usual newsroom order. Most coverage opens with whether discrimination happened; this commentary treats bias as alleged and jumps straight to who absorbs the cost.

What Reuters' framing signals

The word "should" does heavy lifting in that headline. Reuters isn't asking who does pay under current U.S. law, where the answer for most employers is well-settled under Title VII. The commentary asks who ought to pay — a policy question aimed at vendors, employers and regulators at once.

That phrasing matters because vendors of automated screening tools typically draft their contracts to limit liability. HR teams under hiring pressure often sign those contracts without pushing back on indemnification caps. The commentary lands as a growing stack of worker complaints turns that arrangement into a litigation question, not a procurement footnote.

The liability split HR already carries

Under U.S. civil rights law, the employer remains the defendant in a discrimination suit regardless of whether a human manager or an algorithm made the call. Vendor indemnification clauses can shift some financial cost. They rarely shift legal exposure. A worker who sues names the employer, not the vendor, in the initial complaint in the great majority of filings, and the vendor joins the suit later, if at all.

That gap is the space the Reuters commentary walks into. The piece treats the indemnification model as the story, not the algorithm itself. HR leaders reading it should treat it the same way.

What HR teams should pull from this

Three questions worth raising with procurement and legal counsel this quarter:

• Does the AI hiring contract cap vendor liability at fees paid, or does it indemnify the employer against discrimination findings and penalties?
• Can the vendor produce a documented adverse-impact audit on the specific tool and the specific dataset in use, not a generic white paper?
• Does the employer's own audit log survive a litigation hold without surfacing scoring logic the vendor wanted kept proprietary?

The international frame

The Reuters commentary focuses on U.S. liability, but the question it raises cuts across borders. The EU has moved faster than the U.S. on regulating algorithmic hiring tools, classifying them as high-risk and requiring conformity assessments before deployment. U.S. cities and states have followed with their own bias-audit rules, though none match the EU's reach.

These frameworks pull in different directions on who carries the audit duty. HR teams operating in multiple jurisdictions now juggle overlapping rules — and the Reuters commentary lands just as the patchwork grows denser, not clearer.

What to watch next

Reuters' commentary is opinion, not a court ruling or agency action. It lands, though, as EEOC enforcement work, state bias-audit statutes and federal court cases move in parallel. The next ruling against an employer using AI screening will test whether vendor contracts can keep shielding buyers from the Title VII risk they nominally carry. HR leaders should expect that test, not wait for it.