Mercor recruits experts in many fields to train AI models. CBS's 60 Minutes reported in October 2026 that when the big AI companies want a model to improve in a field, they "hire Mercor to make their models smarter." Mercor says its October 2025 funding round valued it at $10 billion. Applicants are interviewed by an AI, on camera.
When we read Mercor's help pages on 7 October 2026, they said one interview serves later applications: "if another application necessitates the same interview, your existing result will be retained." Recordings and an AI-generated transcript are kept "to evaluate your candidacy and to recommend your profile to potential employers," and the person recorded can't have them: "we're not able to share interview recordings or transcripts." Mercor's page on data and AI said profiles are shown to companies unless the candidate switches off a "Looking for Work" setting, and that "Deleting your account does not remove data from companies that may have already accessed it." A feature called Instant Offers shows past interviewees to companies as "prequalified" for roles they didn't apply to.
On training, its pages pointed two ways. The FAQ said interview data "will not be employed to train AI models." The data page listed "Train our internal recommendation models" among its uses, and its terms, last updated on 26 August 2026, take a perpetual licence over what users submit for purposes that include "training models."
On 31 March 2026 Mercor confirmed that an attacker had been inside its systems, through a compromised version of an open-source library. In June it told the people affected that their personal information had been downloaded. WIRED reported that Meta paused its work with Mercor indefinitely.
Seven lawsuits followed and were combined into one case in May 2026. One complaint quotes press reports that the stolen data included "high-definition recordings of candidate interviews, passport and driver's license scans, and facial biometric data used for identity matching." It alleges that applicants "who never performed a single hour of work" for Mercor still had to hand over interview, identity and in some cases biometric data, and that Mercor never showed one plaintiff that his interview video had been deleted. Other complaints allege that Mercor used recorded interviews to train AI models. Mercor disputes the claims, and the combined case has been paused for mediation since July 2026.
What it shows
On a marketplace like this, one recorded interview is kept and used again: shown to other companies, offered for roles the person never applied to, and left with any company that already pulled the profile after the account is closed. The pages that describe this also say the person recorded can't have a copy. A breach then puts the face, the voice and the ID in other hands together, and the lawsuits allege that is what happened here.
For your own record
If you interviewed with Mercor, or with any platform that reuses one interview for many companies, switch off any setting that shows your profile to companies. Ask for a copy of your recording, transcript and scores, and for the names of the companies that have accessed your profile. Ask them to stop using your interview for other roles and for training, and when you're done, to delete it and to pass the deletion on to every company that pulled your profile. Keep their reply: if your recording turns up somewhere later, you'll have what they told you.