Jobs, HR and recruiting: what AI is really doing

What's actually changing

1. 1

   AI runs the first-round interview, not just the resume filter

   How it works. Agentic recruiting platforms now conduct the actual screening conversation, by voice, video, phone or text, before a human recruiter is involved. HackerRank's Chakra hands a candidate a real code repository with an AI assistant in the canvas, asks follow-up questions about why they chose each approach, and, HackerRank says, folds the recruiter screen, take-home test and engineer interview into one session. Workday's Talent Acquisition Agent watches open roles, surfaces qualified applicants and drafts outreach that a recruiter must approve; HeyMilo AI has screened more than one million candidates.

   Why it matters. This decides who a human recruiter ever meets, which matters most for the high-volume, entry-level roles where a first-round human screen used to be the only chance to correct for a bad resume.

   What's overstatedThe agent scores, it does not hire: HackerRank says Chakra scores candidates and leaves the hiring decision to humans, Workday says outreach needs recruiter approval but customers choose which actions run on their own, Workday's agent is available to early adopter customers with general availability planned for early 2027, and Ashby's AI Interviewer was in private beta at its May announcement.

   HackerRankWorkdayHeyMilo AIAshbySapia.aiFika Jobs

   8 receipts

   • HackerRank made its AI interviewer Chakra generally available after it conducted more than 500,000 interviews in testing at companies including Snowflake, Snorkel and Capgemini. After around six months in beta, HackerRank is making Chakra generally available to its customers on Monday. The startup says the AI interviewer has already conducted more than 500,000 interviews during testing, with companies including Snowflake, Snorkel, and Capgemini among those that tried it TechCrunch, 5 Oct 2026 ↗
   • In a Chakra interview the candidate works through a real code repository in a canvas with an AI assistant, and Chakra asks follow-up questions such as why they chose one approach over another. A candidate gets a task involving a real-world code repository, and they work through it in a canvas that includes an AI assistant. As the candidate works, Chakra can use the context of what they are doing to ask follow-up questions, such as why they chose one approach over another TechCrunch, 5 Oct 2026 ↗
   • HackerRank says Chakra is designed to score candidates, and the final hiring decision remains with humans. Ravisankar told TechCrunch that Chakra is designed to score candidates rather than make the final hiring decision, which remains with humans. TechCrunch, 5 Oct 2026 ↗
   • Workday's Talent Acquisition Agent works across Workday Recruiting, HiredScore and Paradox. The Talent Acquisition Agent works across the tools teams already use, including Workday Recruiting, HiredScore, Paradox, native Workday skills, and connected sourcing providers. Workday, 13 Oct 2026 ↗
   • Workday's agent sources in the background, but outreach to candidates still needs recruiter approval before it is sent. Sourcing runs in the background, but outreach to candidates still requires recruiter approval before anything is sent. Workday, 13 Oct 2026 ↗
   • Workday customers decide which actions an agent takes on its own, which need review, and when a person makes the final call. Customers choose which tasks run automatically and how far each agent can go. They decide which actions an agent can take on its own, which need review, and when a person makes the final call. Workday, 13 Oct 2026 ↗
   • du's release says the Sapia.ai partnership brings AI-powered chat interviews to its hiring for corporate and frontline roles. AI-powered chat interviews supporting faster, fairer hiring across corporate and frontline roles Zawya (du press release), 29 Sept 2026 ↗
   • HeyMilo AI has screened more than one million candidates and serves customers including Randstad, WilsonHCG and Neo Financial. The company said it has screened more than one million candidates and serves customers including Randstad, WilsonHCG, and Neo Financial. Pulse 2.0, 17 June 2026 ↗
2. 2

   Candidates fight AI screening with AI, so recruiters retreat to the room

   How it works. Nearly two-thirds of candidates now use AI in applications and most US hiring managers have encountered AI-generated interview answers, so both sides are racing each other: recruiters run AI avatars to conduct first-round interviews, and candidates now send AI avatars or real-time AI coaching to answer them. A Sydney recruiter caught three avatar candidates in a row and disqualified each within minutes.

   Why it matters. When a machine is answering the machine that is screening it, the interview stops measuring anything, and employers are responding by reintroducing mandatory in-person rounds, which strips out exactly the remote and international candidates online hiring was built to include.

   What's overstatedGartner's projection that one in four candidate profiles will be fake by 2028 is a forecast, not a measurement, and detection is often trivial: the Sydney recruiter caught every avatar candidate within about five minutes.

   HireID

   4 receipts

   • A 2026 Greenhouse report found 91% of US hiring managers have encountered or suspected AI-generated answers during online interviews, and 38.5% of 19,368 analysed interviews were flagged for AI-assisted behaviour. A 2026 Greenhouse report found 91% of US hiring managers have encountered or suspected AI-generated answers during online interviews. The Next Web, 8 Aug 2026 ↗
   • Seventy-two percent of recruiting leaders now run at least one in-person interview stage specifically to counter AI-assisted fraud, with Google, McKinsey and Cisco reinstating mandatory face-to-face rounds. Seventy-two per cent of recruiting leaders now run at least one in-person stage specifically to counter AI-assisted fraud, with Google, McKinsey, and Cisco reintroducing mandatory face-to-face rounds. The Next Web, 8 Aug 2026 ↗
   • Across 19,368 live interviews analysed between July 2025 and January 2026, 38.5% of candidates were flagged for AI-assisted behaviour. Across 19,368 live interviews analysed between July 2025 and January 2026, 38.5% of candidates were flagged for AI-assisted behaviour. The Next Web, 8 Aug 2026 ↗
   • HireID's Interview Integrity Platform combines more than 20 live fraud signals with over 98% synthetic-identity detection accuracy in internal testing. combining more than 20 live fraud signals, real-time guidance for interviewers, over 98% synthetic-identity detection accuracy in internal testing Yahoo Finance, 17 Aug 2026 ↗
3. 3

   Layoffs get blamed on AI faster than anyone has proven it caused them

   How it works. Challenger, Gray & Christmas counts the reasons US employers give when they announce cuts. AI is the leading reason year to date, at 120,136 cuts, and hit 40% of May's cuts. Then it faded: in September employers cited AI for 3,961 cuts, the fifth most cited reason. Layoffs.fyi separately counts fifty-four percent of 2026 layoff events citing AI or automation. Both counts track what companies say in a press release, not an audited cause.

   Why it matters. Whether AI is the real cause or a convenient label for ordinary restructuring changes how workers, unions and regulators should respond, and right now the industry's own experts do not agree with each other.

   What's overstatedOpenAI's own CEO says companies are "AI washing" their layoffs, Apollo's chief economist says he sees zero evidence of job losses because of AI in ADP's employment report, and a Lightcast economist says the labor data does not show AI replacing workers en masse.

   Challenger, Gray & ChristmasLayoffs.fyiLightcastOracleAmazon

   5 receipts

   • Challenger says AI was the fifth most cited reason for US job cuts in September 2026, with 3,961 cuts, while remaining the leading reason year to date at 120,136. Artificial Intelligence was the fifth-most cited reason with 3,961 cuts in September, about 9% of the month's total. So far this year, AI has been cited in 120,136 job cut announcements, approximately 21% of all cuts, and it remains the leading reason year-to-date. Challenger, Gray & Christmas, 1 Oct 2026 ↗
   • Challenger says AI accounted for 40% of 97,006 US job cuts in May 2026, the highest monthly total since it began tracking AI as a layoff reason in 2023. AI accounted for 40% of 97,006 job cuts by US-based employers in May, the highest monthly total since Challenger began tracking AI as a reason for layoffs in 2023. Business Insider, 4 June 2026 ↗
   • Apollo Global Management's chief economist says he sees zero evidence of job losses because of AI, citing the ADP National Employment Report. Apollo Global Management's chief economist Torsten Sløk wrote last week that he sees "zero evidence of job losses because of AI," citing the ADP National Employment Report. Business Insider, 4 June 2026 ↗
   • A Lightcast economist says the labor data does not show AI replacing workers en masse. There are narratives currently that note AI is "replacing" workers en masse. Simply put, we are not seeing that in the labor data, HRD America, 19 Oct 2026 ↗
   • Layoffs.fyi counts fifty-four percent of 2026 layoff events citing AI, automation or machine learning. Fifty-four percent. That is the share of layoff events in 2026 where companies have explicitly pointed to artificial intelligence, automation, or machine learning as a contributing factor, according to Layoffs.fyi International Business Times UK, 26 Aug 2026 ↗
4. 4

   When an algorithm decides who gets fired, the lawsuits follow

   How it works. Twenty-six current and former Meta employees are suing, alleging an internal AI system scored and ranked staff for the May 2026 layoff without adjusting for protected medical or parental leave, disproportionately selecting employees who were on leave, including one scientist two days before she gave birth. Separately, a federal judge let Mobley v. Workday proceed over claims Workday's AI screening tool discriminated by disability, race, age and gender across roughly 1.1 billion rejected applications.

   Why it matters. These are the first major tests of whether the vendor whose model made the ranking, not just the employer who used it, can be held liable when an algorithm's output harms a protected class.

   What's overstatedWorkday says its technology "looks only at job qualifications, not protected traits," and Meta has not admitted the AI system made the calls; both cases are live allegations, not proven verdicts.

   MetaWorkday

   4 receipts

   • Meta's AI systems allegedly targeted employees with disabilities or on medical or parental leave in its May 2026 layoff. The company's AI systems targeted employees with disabilities or on medical or parental leave, the lawsuit claims. Gizmodo, 15 July 2026 ↗
   • Employees who took protected leave were disproportionately selected for layoff by a scoring system that did not account for that leave. employees who took protected leaves were disproportionately selected for layoff, based on scoring that not only failed to account for their protected leaves HRD America, 15 July 2026 ↗
   • Workday's own court filings disclosed that approximately 1.1 billion job applications passed through its screening system during the period at issue in the Mobley case. Workday's own court filings disclosed that approximately 1.1 billion applications had passed through its system during the period at issue. Tech Times, 23 June 2026 ↗
   • A federal judge ruled Workday must face claims its AI-powered screening tools discriminated against applicants by disability, race, age and gender. must face claims its artificial intelligence (AI)-powered screening tools discriminated against job applicants based on disability, race, age, and gender HRD America, 23 June 2026 ↗
5. 5

   Hiring-AI rules: Colorado gutted its law, Connecticut's just switched on

   How it works. Connecticut's AI Responsibility and Transparency Act took effect October 1, 2026: employers using high-risk AI that makes or substantially informs a hiring, promotion or firing decision must run impact assessments, tell applicants, and let them request human review where feasible. It puts that burden on the employer, not the vendor that built the tool. Colorado went the other way, repealing its pre-deployment bias-testing law after a lawsuit from xAI and replacing it with SB26-189, which from January 2027 brings in developer documentation, notice to applicants, three-year compliance records and a right to request human review, and the EU pushed its high-risk check for recruitment AI to December 2027.

   Why it matters. An employer that buys an AI screener now carries the compliance burden itself in Connecticut and owes applicants notice in Illinois, while the toughest pre-use bias rules in Colorado and the EU were cut or delayed before they bit.

   What's overstatedA Stanford study of 4 million job applications found an AI hiring tool can pass a bias audit at the aggregate level while still systematically screening out Black and Asian candidates role by role.

   ConnecticutColoradoxAIIllinois Department of Human RightsWorkdayHireVue

   16 receipts

   • Connecticut's AI Responsibility and Transparency Act took effect October 1, 2026, and puts the compliance burden on the employer that deploys an AI tool, not its vendor. Even if you purchased or licensed an AI system from a third-party provider, the CAIA places the compliance burden on the deployer, not the vendor that built the technology. Employment Law Letter (Shipman & Goodwin), 17 Sept 2026 ↗
   • Connecticut's law covers high-risk AI systems, those used to make or substantially inform consequential decisions about people. The CAIA regulates the use of “high-risk” artificial intelligence (“AI”) systems, meaning those used to make or substantially inform consequential decisions about people. Employment Law Letter (Shipman & Goodwin), 17 Sept 2026 ↗
   • Connecticut's law requires a right to appeal, letting individuals request human review of an AI-assisted decision where feasible. Offer a right to appeal, allowing individuals to request human review of an AI-assisted decision where feasible. Employment Law Letter (Shipman & Goodwin), 17 Sept 2026 ↗
   • xAI sued in federal court on April 9, 2026 to block Colorado's original AI Act, SB 24-205. On April 9, 2026, Elon Musk's artificial intelligence company xAI filed a federal lawsuit in the US District Court for the District of Colorado — xAI v. Weiser, No. 1:26-cv-01515 — seeking to block SB 24-205 on constitutional grounds. Tech Times, 23 June 2026 ↗
   • Colorado's governor signed SB26-189 on May 14, 2026, repealing and replacing the original AI Act. Governor Jared Polis signed Senate Bill 26-189 on May 14, 2026, formally repealing and replacing the original law forty-seven days before it was scheduled to take effect. Tech Times, 23 June 2026 ↗
   • SB26-189 removes the original act's duty of care, risk management programs, impact assessments and annual reporting to the Colorado Attorney General. Senate Bill 26-189 eliminates all of that. The duty of care is gone. Risk management programs are gone. Impact assessments are gone. Annual reporting to the Colorado Attorney General is gone. Tech Times, 23 June 2026 ↗
   • From January 1, 2027, SB26-189 requires developers of automated decision tools that materially influence consequential decisions to give deployers technical documentation, including instructions for appropriate use and human review. The act requires the developer of an ADMT (developer) that is used to materially influence a consequential decision (covered ADMT), starting January 1, 2027, to provide a deployer of a covered ADMT (deployer) with technical documentation describing the covered ADMT's intended uses, categories of training data, known limitations, and instructions for appropriate use and human review. Colorado General Assembly (summary of the act as enacted), 14 May 2026 ↗
   • SB26-189 requires both developers and deployers to keep the records that show compliance for at least three years. Both developers and deployers are required to retain records necessary to demonstrate compliance with the act for at least 3 years. Colorado General Assembly (summary of the act as enacted), 14 May 2026 ↗
   • SB26-189 requires deployers to give consumers clear and conspicuous notice at the point of interaction with a covered automated decision tool. The act establishes consumer notice requirements, mandating that deployers provide clear and conspicuous notice to consumers at the point of interaction with a covered ADMT. Colorado General Assembly (summary of the act as enacted), 14 May 2026 ↗
   • After an adverse outcome, SB26-189 requires the deployer to describe the tool's role in plain language within 30 days. A deployer is required to provide a consumer with a plain language description of a covered ADMT's role within 30 days after the covered ADMT makes a consequential decision that results in an adverse outcome for the consumer. Colorado General Assembly (summary of the act as enacted), 14 May 2026 ↗
   • SB26-189 gives consumers the right to correct factually incorrect personal data a covered tool used and to request meaningful human review and reconsideration after an adverse decision. Consumers have the right to request personal data and correction of factually incorrect personal data used by a covered ADMT. The act also grants consumers the right to request meaningful human review and reconsideration following a covered ADMT making a consequential decision resulting in an adverse outcome. Colorado General Assembly (summary of the act as enacted), 14 May 2026 ↗
   • Illinois House Bill 3773 took effect January 1, 2026, requiring employer notice whenever AI influences a covered employment decision. House Bill 3773 took effect on January 1, 2026, and the Department of Human Rights is tasked with adopting rules to implement and enforce these new requirements JD Supra, 27 Feb 2026 ↗
   • JD Supra says Illinois' draft rules show how the state intends to implement the notice provisions of House Bill 3773. Nevertheless, the draft rules offer valuable insight into how the IDHR intends to implement the notice provisions of House Bill 3773. JD Supra, 27 Feb 2026 ↗
   • NYC's automated-hiring law only applies to software that substantially assists or replaces a hiring decision, leaving a loophole when a human is still part of the process. The law, however, only applies to software that "substantially assists" or replaces decision-making, creating a loophole for cases where humans are part of the process. The Guardian, 19 Aug 2026 ↗
   • The EU's Digital Omnibus pushed the high-risk conformity assessment deadline for recruitment AI, resume screening and interview assessment to December 2027. Recruitment AI — résumé screening, candidate filtering, interview assessment, job ad targeting — is explicitly listed in Annex III and carries, under the Omnibus, a December 2027 deadline. Tech Times, 10 July 2026 ↗
   • A Stanford study of 4 million job applications found an AI hiring tool can pass a bias audit in aggregate while still screening out Black and Asian candidates for specific roles. A study of 4 million job applications found that an AI hiring tool can pass a bias audit at the aggregate level while still systematically screening out Black and Asian candidates for specific roles. HRD America, 19 June 2026 ↗
6. 6

   Employers watch workers with AI that mostly does not prove it works

   How it works. The global employee-monitoring market, tracking location, keystrokes, phone calls and even tone of voice, is on track to exceed $4 billion in 2026. A review of the research and a meta-analysis of more than 23,000 subjects both found essentially no link between monitoring and performance, while monitoring reliably raised workplace stress.

   Why it matters. Employers are spending billions on AI surveillance to raise productivity, with peer-reviewed evidence showing a real cost, more stress and workarounds, but no reliably measured benefit.

   What's overstatedVendors sell monitoring as a productivity tool, but the closest thing to a controlled result, a study of remote Upwork workers, found output only fell when top performers had monitoring removed without being told why, and held steady when they were simply told the reason, an argument for management transparency, not more surveillance software.

   Amazon Web Services

   3 receipts

   • The global market for employee-monitoring technology is expected to exceed $4 billion in 2026, driven largely by major North American employers. the global market for employee-monitoring technology is expected to exceed $4 billion in 2026, largely due to major companies in North America Scientific American, 8 Sept 2026 ↗
   • A review of studies on employee monitoring found little evidence that electronic surveillance improves worker performance. When König reviewed studies on employee monitoring in the 2025 issue of the Annual Review of Organizational Psychology and Organizational Behavior, he found little to suggest that electronic surveillance improved performance. Scientific American, 8 Sept 2026 ↗
   • A meta-analysis covering more than 23,000 subjects found no correlation between the level of workplace monitoring and employee performance. Ravid conducted a 2022 meta-analysis that combined the results of studies involving more than 23,000 subjects in laboratory experiments and workplace settings Scientific American, 8 Sept 2026 ↗

Who's doing it

11 companies to know, 5 with a round we can source. Newest first where dated.

Search these companies in the startup database ↗ · Explore all research tools

Your move

Our read, built on the receipts above.

What's next

• Jan 2027 Workday plans general availability of its Talent Acquisition Agent autopilot (company says early 2027) newsroom.workday.com ↗
• Jan 2027 Colorado's replacement AI law (SB26-189) starts applying: from January 1, 2027, developers of automated decision tools must give deployers technical documentation, including instructions for appropriate use and human review leg.colorado.gov ↗
• Dec 2027 The EU AI Act's high-risk conformity assessment deadline for recruitment and employment AI systems arrives techtimes.com ↗