Leadership1 distinct publisher3 min readUpdated
Chair Mike Reed disclosed an audience-data partnership on an August 6 earnings call. Within days, more than 800 NewsGuild-CWA members demanded the company end it.
The Board Room · Leadership desk

Compiled by The Board RoomSomething wrong?How this is made
On an August 6 earnings call, USA Today Co. chair Mike Reed told investors the company is working with Palantir to build a common intelligence layer on top of its audience data, aimed at faster monetization across subscriptions, advertising and commerce, per a Yahoo account cited by Forbes [1]. The journalists who produce that audience found out the same way investors did [2], and within days more than 800 unionized journalists and media workers represented by the NewsGuild-CWA demanded the company end the deal immediately, according to Common Dreams as cited by Forbes [3].
That sequence is the whole story. A revenue project that would have been a vendor selection in any other quarter is now a bargaining-table item, and the escalation took roughly nine days from disclosure to public revolt [4].
The asset in question is first-party data: the behavioral record a publisher collects directly as readers move through its sites, including what they open, how long they stay, what they click, and which subscription offers they refuse or accept [5]. USA Today Co. operates more than 200 news outlets, so those signals aggregate into behavioral profiles at national scale [6]. Palantir was retained to unify that view and predict what each reader does next [7].
The guild's objection, according to the Forbes account, is not the analytics [8]. It is the counterparty. Palantir holds a thirty million dollar contract with Immigration and Customs Enforcement [9], and its software has powered surveillance and immigration operations that USA Today newsrooms cover as a core beat [10]. The union argues that partnering with a major subject of its own coverage creates an inherent conflict of interest and raises concerns about the data security of readers [11]. Guild members also note that journalists have been assaulted and arrested while covering immigration enforcement, which makes the choice of vendor personal rather than procedural [12].
None of this makes USA Today Co. an outlier on the technology. Palantir has struck similar deals with Axel Springer and Thomson Reuters [13], the company says it complies with applicable privacy laws and holds vendors to the same standard [14], and nearly every major publisher already runs predictive AI-driven analytics on its audience [15]. What separates this one, in the Forbes framing, is that readers learned their behavior was being modeled by a contractor from an earnings call rather than a privacy notice, and the employees closest to those readers were not told beforehand [16].
The technical mitigations are not exotic. Forbes notes that differential privacy, federated learning, on-device modeling and synthetic data all allow a publisher to predict reader behavior without assembling identifiable profiles in a central warehouse, and argues that what is missing is any public signal that such methods are in use here [17]. The column's operator lessons are equally unglamorous: disclose before the earnings call, vet the partner's whole business rather than the licensed product, inventory what you actually collect, and give customers a working exit [18].
Watch whether management issues a reader-facing description of what the partnership touches, since the absence of one is the gap Forbes identifies [17]. Watch whether the guild's demand moves into formal bargaining, given that more than 800 represented workers have already signed on [3]. And watch Axel Springer and Thomson Reuters for the same internal question, since they bought from the same vendor [13].
Follow any of these and your For You feed starts watching them — no settings page required.
Ranked by verification strength, evidence, and original report placement.
On an August 6 earnings call, USA Today Co. chair Mike Reed told investors the company is working with AI and analytics firm Palantir to build a common intelligence layer on top of its audience data, aimed at driving faster monetization across subscriptions, advertising and commerce.
The journalists who work at the company found out about the Palantir arrangement the same way investors did.
Within days of the earnings call, more than 800 unionized journalists and media workers represented by the NewsGuild-CWA demanded the company end the deal immediately.
First-party data is the behavioral record a publisher collects directly as readers move through its sites: every article opened, time spent, clicks, and which subscription offers are ignored or accepted; it is called first-party because the publisher gathers it rather than buying it from a broker.
USA Today Co. operates more than 200 news outlets, so its reader signals add up to detailed behavioral profiles at national scale.
Palantir was hired to unify the company's view of reader data and predict what each reader will do next.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Single aggregated account, no primary documents
One contributor article carries the entire cluster and explicitly relays the two load-bearing facts from other outlets (Yahoo for the earnings call, Common Dreams for the union demand). No transcript, union statement, contract, privacy notice or company response is supplied, and the text contains copy errors including a garbled counterparty name, which further limits verifiability.
Announced engagement, undocumented deployment
Adoption evidence is limited to the fact of an announced partnership and an unquantified assertion that similar deals and predictive audience analytics exist elsewhere. There is no disclosed rollout scope, data volume, integration status, timeline or measurable outcome, and the countervailing observation is organized internal resistance rather than expansion.
Framing outruns the documented record
The article opens by calling this 'the most heated AI data fight in America media' and headlines a 'newsroom revolt' while resting on two secondhand facts, and it presents privacy-preserving alternatives as an available fix with no visibility into the actual architecture. The overstatement is moderate rather than severe because the piece explicitly concedes the deal is not an outlier and confines its own contribution to prescriptive lessons.
All named parties have stakes in the framing
Every actor in the story is interested: the company disclosed the partnership to investors as a monetization accelerant, Palantir's on-record position is a compliance assurance, the guild is pressing an active bargaining and reputational demand, and the article itself is a contributor piece that converts the episode into an executive advisory framework. These incentives are visible on the face of the source rather than inferred.
Low — one secondhand publisher, no rebuttal
The core sequence (August 6 disclosure, then a large union demand) is internally coherent and dated, which supports modest confidence, but everything rests on one aggregating publisher with no company response, no primary documents, no independent confirmation of the ICE contract figure, and demonstrable transcription errors.
invest
The 81% Problem: AI's Star CEOs Are Polling Badly With The People They Need To Hire1 distinct publisher
invest
Chainalysis sues over ICE's $94.7M award to TRM Labs, and the filing is sealed1 distinct publisher
product
Washington's secret AI test is coming for open weights, and release dates go with it2 distinct publishers
product
Cinemas, classrooms and ICE: smart glasses now need a venue-policy contingency1 distinct publisher
Distinct publishers with included, body-backed reporting in this cluster.
1 article · August 15, 2026