Eightfold AI lawsuit: alleged hidden scores and what applicants can check

U.S. coverage — October 3, 2026

The Eightfold AI lawsuit challenges whether AI-generated applicant evaluations should carry the transparency protections that apply to covered employment consumer reports. It does not establish that Eightfold discriminated against applicants or that every rejection at an employer using Eightfold was caused by AI.

In Kistler v. Eightfold AI Inc., Erin Kistler and Sruti Bhaumik allege that Eightfold assembled information about applicants, scored them without meaningful access to the resulting reports, and failed to comply with federal and California consumer-reporting laws. Eightfold denies the allegations. For candidates, the useful next step is to identify what happened to a specific application—not infer a hidden score from repeated rejections. Complaint; Eightfold’s response reported by the Guardian.

What the lawsuit alleges

The complaint, dated January 20, 2026, describes a process that goes beyond storing or parsing a résumé:

• Eightfold allegedly combines application information with employer data and third-party online information.
• Its tools allegedly generate inferences about applicants’ skills, characteristics and career prospects, then rank their “likelihood of success” from 0 to 5.
• Employers allegedly use those evaluations to prioritize candidates, while applicants cannot meaningfully inspect or correct the information before it affects hiring.

These are the plaintiffs’ allegations, not findings about what happened to every applicant. The complaint asserts claims under the Fair Credit Reporting Act (FCRA), California’s Investigative Consumer Reporting Agencies Act (ICRAA) and California’s Unfair Competition Law. Its central legal theory is that Eightfold operates as a consumer reporting agency and supplies covered employment reports—not simply that AI screening is unfair. Complaint, paragraphs 1–14 and claims for relief.

Eightfold’s current product page says its recruiting suite uses multimodal large language models to evaluate more candidates and “look beyond résumés.” That is a vendor description of capability, not proof that a particular applicant received a score, was automatically rejected or suffered discrimination. Eightfold Talent Acquisition.

Where the case stands—and who it seeks to cover

Eightfold removed the case from Contra Costa County Superior Court to the U.S. District Court for the Northern District of California on March 2, 2026. The federal case number is 3:26-cv-01768. Eightfold filed a motion to dismiss on April 20; briefing followed in June and July. The court vacated the scheduled hearing on July 24.

The available public docket lists entries through August 12, 2026, and does not show a ruling on the motion to dismiss or a class-certification order. That record may lag the court’s complete file; it does not establish that no later order exists. Public docket.

The original complaint proposes:

• A nationwide class of U.S. residents who applied to U.S. jobs and were subjected to Eightfold’s evaluation tools, within applicable limitation periods.
• A California class of California residents whose job applications were subjected to those tools, within applicable limitation periods.

Those are proposed definitions, not a confirmed eligibility list or promise of compensation. Merely applying to a company that uses Eightfold does not establish that the relevant tools evaluated your application. Complaint, paragraphs 115–124.

What applicants can do now

1. Preserve the application record. Keep the job description, requisition number, submitted résumé, application date, confirmation, rejection message and any notices or authorization forms. Save the application portal address and screenshots showing a vendor’s involvement. These document the application flow; they do not prove why you were rejected.

2. Ask the employer about that application. A focused request is more useful than asking whether the company “uses AI” generally:

For my application to [role and requisition number] on [date], did Eightfold or another third-party system generate a score, ranking, profile or recommendation? If so, please identify the provider, the information used and any available process for accessing or correcting my records. Was the output used to determine whether my application advanced?

This is a request, not a guarantee of disclosure or a right to a new interview.

3. Send privacy requests to the appropriate organization. Eightfold’s May 2026 notice describes its role when processing customer data as a service provider and directs candidates to the employer for access, correction or deletion requests concerning that data. It separately provides a privacy webform for exercising applicable rights concerning data it holds in its own capacity. This is Eightfold’s stated privacy-request process, not a court determination of its status under consumer-reporting law. Available rights depend on applicable law. Ask specifically about application data, third-party information and derived evaluations—not only the résumé you uploaded. Eightfold Privacy Notice, sections B, C and D.

4. Distinguish a privacy notice from an FCRA process. If a covered consumer report is used for employment, the FTC explains that employers generally must provide a standalone written disclosure and obtain written permission before obtaining it. Before an adverse decision based on the report, they must provide the report and a summary of FCRA rights; an additional notice follows the adverse action. Whether Eightfold’s disputed evaluations qualify is the issue the lawsuit raises. FTC employer guidance.

For the classification question, see when an AI résumé score may count as a background report. Applicants considering a claim should consult qualified counsel rather than assume the proposed class preserves their individual deadlines.

What HR teams should check

Map the actual workflow: which data enters the tool, what it infers, what recruiters see, and whether scores affect advancement. Ask counsel to assess consumer-reporting obligations separately from privacy and discrimination requirements. A general AI notice is not a substitute for a covered-report process, and a human clicking the final button does not tell you how much the ranking influenced selection.

Use an AI hiring vendor validation checklist to document these questions before deployment or renewal. Candidates need a route to raise errors, and employers need records of what their tools contributed to the decision.

Informational only—not legal, HR or employment advice. Requirements and remedies depend on the jurisdiction and facts.