Technology & innovation
AI-based hiring tools showed racial bias against Black and Asian job applicants in a substantial share of job openings, according to a new Stanford study examining more than 4 million job applications.
15% of Asian applicants and 26% of Black applicants applied to jobs where the AI screening tool appeared biased against their racial group, recommending them at less than 80% the rate of the leading group.
Researchers estimated that 40,000 more applications from Asian and Black candidates would have been recommended had racial groups been selected at equal rates.
It was surprising that AI systems that use game-based assessment to rank people were still biased against Black and Asian applicants."
The study also found evidence of "systemic rejection," where 4% of applicants who applied to 10 positions were rejected by the AI tool for all 10, a higher rate than would occur if companies were making independent decisions.
About 90% of employers use AI to some extent in hiring, but researchers say independent study of these tools remains rare because hiring data is typically kept private.
Read more via Stanford
CTO confidence in scaling AI has fallen to 48% in 2026 from 82% in 2024, even as AI adoption accelerates, according to Akkodis' recent survey of 500 CTOs.
Just 44% of CTOs believe leadership teams have sufficient AI understanding, and only 46% report established frameworks for responsible AI.
Top barriers to scaling AI include a lack of in-house technology skills (32%), uncertainty around ROI (31%), and a lack of urgency (27%).
Only 21% of CTOs report workforce reduction due to AI, while 50% report changes in required skills and 49% report shifts in day-to-day activities.
For the first time, CTOs cited innovation, not efficiency, as the primary driver of digital investment.
What we're seeing now is not a slowdown in AI adoption, but a moment of realism. Organizations are moving beyond experimentation and encountering the reality of scaling AI across complex environments."
Read more via GlobeNewswire
A venture capitalist's viral post is drawing attention to growing tension between engineers who have embraced heavy AI use and those tasked with fixing the resulting code, as employers increasingly mandate AI tool adoption.
VC experts say "most software engineers are facing an identity crisis bordering on depression" as experienced engineers are left cleaning up AI-generated code.
Some employers, including Meta, now factor employees' AI usage into performance reviews.
"Tokenmaxxing," a term for excessive AI usage as a point of pride among engineers, has gained traction this year.
Separate research has documented "workslop," or low-quality AI output that creates the appearance of productivity while shifting cleanup work onto colleagues.
Read more via Futurism
Deloitte has deployed a new suite of autonomous AI agents to its global audit teams, part of the firm's $3 billion, five-year investment in AI capabilities.
The new agents can scan media reports for emerging threats, draft staff memos, and help sift through transactions and client documents.
Deloitte says the tools will speed up response times to client questions and help auditors evaluate disclosure and regulatory requirements.
The rollout follows similar moves by EY and KPMG, which have also introduced autonomous AI agents into global audit work this year.
Deloitte says the goal extends beyond efficiency to faster document turnaround, potentially freeing auditors for broader analysis.
We expect to see the time from when the documents come in from a client to when they get populated in working papers to start to accelerate. It means they can execute their tasks more quickly in the course of the day and hopefully they can then ask broader questions as well."
Read more via Bloomberg Tax
Companies making the largest investments in generative AI grew employment by roughly 10% over the two years following adoption, according to a new study from Ramp and Revelio Labs that tracked 21,559 U.S. firms.
Entry-level headcount grew even faster than overall headcount, rising 12% among high-intensity AI adopters.
Low-intensity AI adopters saw no statistically significant change in employment.
Employment gains were broad across roles, including engineering, sales, administration, and customer service, and emerged gradually over time.
AI adopters tend to already be larger, more technical, and faster-growing firms than non-adopters, meaning the gains are concentrated among companies already positioned to grow.
Read more via Ramp
A new survey from Hudson Talent Solutions finds that while AI is improving recruiting efficiency, most talent professionals still don't fully trust it, underscoring the continued need for human oversight in hiring.
94% of respondents say AI-generated insights are sometimes or often inaccurate, and just 7% say they trust AI more than employees for accurate, ethical business insights.
54% say a combination of people and AI delivers the best results in recruiting.
62% say AI has had a positive impact on job satisfaction, but 58% say their organization doesn't offer on-the-job AI training.
34% of respondents say AI proficiency is now a necessary skill for all job candidates, and 44% say listing it on a resume would help land a job.
AI is changing what is possible in recruitment, but it does not replace the need for experienced recruiters, hiring managers, and strategic talent advisors."
REI's AI ad grew an extra pair of handlebars: An Instagram ad for REI's Van Rysel EDR AF bike went viral after Meta's AI personalization tool altered the image to show handlebars at both ends. Meta had auto-enrolled REI in the generative ad tool without the retailer's request, and REI has since dropped out of the program. The model featured in the ad said she didn't recognize the altered image of herself, despite having been hired for the original photo shoot. (TechSpot)
Pentagon and OPM launch "War Force" to recruit AI and software engineers: The Office of Personnel Management and the Department of War have launched a recruiting initiative connecting experienced software engineers with technical roles supporting national security missions. The program, part of OPM's broader Tech Force initiative, is targeting candidates with backgrounds in AI and machine learning, automation, data systems, and large-scale software platforms, who may work directly alongside operational units rather than staying in headquarters roles. Applicants must be U.S. citizens eligible for a Secret or Top Secret clearance. (ClearanceJobs)
Mental health chatbots are booming, with little research behind them: More than 100 AI chatbots now market themselves as tools for depression, anxiety, and addiction support, even as researchers find inconsistent safety performance. A Stanford study found one bot responded to a subtle suicide risk cue by listing bridge heights rather than flagging the danger. Companies like Talkspace and Headspace have built their own bots with therapist oversight and guardrails such as session time limits, and CMS is piloting a reimbursement model for some of these tools starting in July. Critics, including the American Psychological Association, say the tools remain unregulated by the FDA even as several states move to regulate their use. (The Wall Street Journal)
