In January 2026, a class action lawsuit revealed that one AI hiring platform had secretly scored over one billion workers — assigning each a zero-to-five rating — and discarded low-ranked candidates before a single human ever saw their application. Most HR teams deploying AI screening tools have no idea this is how the process actually works.

The lawsuit against Eightfold AI is not just a legal curiosity. It is the most visible crack in a system that has quietly become the infrastructure of modern hiring — and it signals that a broader reckoning for algorithmic screening is well underway.

The Eightfold AI Lawsuit: What Actually Happened

The January 2026 class action was brought by former EEOC chair Jenny R. Yang and the nonprofit Towards Justice. Their allegation: Eightfold AI scraped personal data on over one billion workers from social media profiles, browsing activity, and other public sources, then built individual "dossiers" scoring each person's likelihood of success in a given role — and sold those reports to employers without candidates ever knowing they existed.

The legal hook is not bias. It is the Fair Credit Reporting Act (FCRA). The plaintiffs argue that Eightfold was compiling consumer reports — a legally defined category that triggers strict disclosure and consent requirements — without following mandatory procedures. Eightfold denies the allegations, stating its platform "operates on data intentionally shared by candidates or provided by our customers."

Whether or not the FCRA claim succeeds in court, the core allegation is damning on its own terms: millions of candidates were evaluated, scored, and discarded by an algorithm they never consented to, based on data they never knew had been collected.

The Workday Case Set the Legal Stage

Eightfold's lawsuit did not arrive in isolation. It follows Mobley v. Workday, the landmark case where a federal court ruled that Workday acted as an "agent" of the employers using its automated screening tool — not merely a software provider. That ruling allowed class action certification potentially covering millions of applicants over 40 who were filtered out by Workday's AI.

This matters enormously. The traditional legal shield for enterprise software vendors — "we just provide the tool, employers make the decisions" — is crumbling in court. When a platform processes applications and surfaces ranked candidates, courts are increasingly treating that as a hiring decision, not a neutral technology service.

Employers who assumed their AI vendor absorbed the legal risk of algorithmic screening now face a different reality. The liability follows the hiring outcome, not the software contract.

The 60-80% Problem: Who Is Actually Deciding

Here is what makes these lawsuits so significant for HR teams beyond the legal headlines. Research from 2025 indicates that AI tools handle between 60 and 80 percent of the early hiring funnel at large employers. In practice, the algorithm decides who gets seen. A candidate rejected at the Eightfold stage would never know they had been scored, let alone what score they received or why.

An internal audit leaked during the Mobley v. Workday proceedings found that the AI screening tool rated older male candidates systematically higher than equally qualified female and younger candidates — because the training data reflected historical hiring patterns. The AI had not invented the bias. It had faithfully reproduced and then automated it at scale, across millions of applications, at a speed no human auditor could track in real time.

"The algorithm did not introduce bias into hiring. It industrialised the bias that was already there."

This is the AI hiring bias problem in its most precise form: not a rogue model making obviously wrong decisions, but a well-functioning model optimising for a flawed historical signal — and doing so invisibly, consistently, and at enormous scale.

"The Algorithm Did It" Is No Longer a Legal Defence

The EEOC has been unambiguous: employers cannot outsource legal accountability to their AI vendors. If an algorithmic screening tool disproportionately filters out protected groups — regardless of intent, regardless of whether the vendor built the model — the employer is exposed under Title VII, the ADEA, and the ADA.

This creates a genuine liability problem for HR teams that have deployed AI screening without full transparency into how the models actually work. Many enterprise contracts give employers access to a dashboard and a ranked list of candidates. They do not give employers access to the underlying model logic, training data, or algorithmic decision criteria.

In legal terms, companies are making consequential employment decisions via black box — and courts are now treating that as the employer's problem, not the vendor's.

New York City's Local Law 144 already requires annual independent bias audits for any AI hiring tool used by employers within the city. Colorado, Maryland, and Illinois have similar legislation in progress at the state level. The compliance landscape is tightening fast, and most legal teams are not keeping pace.

Three Questions Every HR Team Should Ask Their AI Screening Vendor Now

This is not an argument to abandon AI hiring tools. Used transparently, with human oversight and regular auditing, algorithmic screening can add genuine value. The problem is deployment without due diligence — and that is where most organisations currently sit.

Three questions that should be put to every AI screening vendor before the next renewal:

Does your tool produce a "consumer report" as defined under the FCRA? If the platform compiles information from public data sources — social media, browsing behaviour, professional databases — to evaluate candidates, the answer may be yes. Compliance obligations, including disclosure and consent requirements, follow automatically.

Has your screening model been independently audited for disparate impact? Vendor-provided compliance documentation is not the same as an independent bias audit. Ask for the methodology, the protected class breakdowns, and the third-party auditor's name. If the vendor cannot provide this, that is the answer.

What data is being collected on candidates who did not apply directly through your ATS? If the tool sources external data on candidates — as Eightfold allegedly did — your organisation may have disclosure and consent obligations you are currently unaware of, and exposure you have not accounted for.

The Bottom Line

The Eightfold lawsuit is not an isolated event. It is the leading edge of a legal reckoning that has been building since AI hiring tools went mainstream, and the Workday case has already shown that courts are willing to hold both vendors and employers accountable for algorithmic outcomes.

The companies most exposed are those that deployed AI screening quickly, without auditing the models, understanding the data inputs, or tracking disparate impact on protected groups. In 2024, that described the majority of enterprise HR teams. In 2026, with class actions in motion and regulators sharpening their focus, it is no longer a defensible posture.

AI hiring algorithms can be powerful tools. But "the algorithm decided" is not a hiring strategy — and courts are making clear it is not a legal defence either. HR leaders who wait for a lawsuit to understand how their screening tools actually work will find out the hard way.

Sharingan AI tracks legal and technical developments in AI recruitment technology so hiring teams have the independent analysis they need to make informed decisions — not just the vendor pitch.