Technology & innovation
A Highwire survey of more than 300 U.S. managers and more than 700 individual contributors found employees are mostly in the dark about it.
Only 16% of employees say they were told AI played a role in their most recent review.
54% of employees found AI-generated feedback more specific and actionable, while 34% said it became more generic.
Two-thirds of employees say they change their behavior to protect themselves, such as avoiding a reasonable risk or holding back an idea.
AI can help you write a faster review, but it doesn’t replace human judgment, context, nuance and empathy.”
Read more via HR Dive
Oyster’s 2026 Global AI Talent Report draws on a survey of more than 1,000 HR professionals and an analysis of more than 15,000 job descriptions.
Among companies that added AI requirements to existing roles in the past year, 62% changed neither title nor pay.
Among companies that added AI skill requirements to non-technical roles, 46% have not formally communicated the change.
53% of companies prefer to train existing employees when AI skills are missing, while 14% primarily hire externally.
AI skill requirements in roles hired through Oyster have tripled since 2022, from 5% to 15%.
Read more via Business Wire
NPR reports that workers in India wear cameras to record their hand movements as training data for AI robots, and some say they were never told.
Textile worker Shoba Kirubakaran, who earns about $90 a month, says no one told her the footage would be used to train robots.
Other workers are paid about $2.60 an hour to record themselves.
Developers need millions of hours of what the industry calls egocentric data, filmed from the worker’s point of view.
Some experts say robots are unlikely to be deployed at scale in India soon because they are expensive and power hungry.
Humans are cheaper to run than robots.”
Read more via NPR
Ars Technica looks at how the Amazon’s growing robot fleet is changing warehouse work.
Amazon says fulfillment centers with advanced robotics need about 30% more employees in reliability, maintenance, and engineering roles.
A 2025 study in ILR Review found robotic warehouses had 40% fewer severe injuries but 77% more less severe injuries tied to repetitive motion.
Amazon says its global recordable incident rate has improved 43% since 2019.
The New York Times reported last year on internal documents suggesting robots could reduce projected U.S. hiring by more than 160,000 by 2027, which Amazon said reflected one group’s views.
Read more via Ars Technica
Elon University’s Imagining the Digital Future Center surveyed 1,158 U.S. adults in August, including 775 workers.
68% expect most jobs to pay too little for a decent lifestyle.
62% expect most available jobs to be dull and unfulfilling.
66% of workers say they are already comfortable using AI at work or would welcome it with more training.
Only 11% of workers are highly confident government agencies would help if AI significantly affects their jobs.
Telling workers to adapt rings hollow when human systems remain rooted in the past and do not invest in the human future.”
Read more via Elon University
CNN reports that China’s AI buildout is moving quickly with little public opposition, though the country trails the U.S. on chips and capital.
The U.S. had roughly 10 times more AI data centers than China as of last year, according to Stanford research.
Nearly 90 data centers are slated to be built in Ulanqab, a city of fewer than two million people.
Beijing announced a $560 billion plan last month to scale up computing and information infrastructure over five years.
Listed Chinese companies will double their AI infrastructure investment this year to about $139 billion, compared with $800 billion from U.S. peers, according to Rhodium Group.
If you look at the overall production cost of building a data center, chips are the most important factor, and they just don’t have enough.”
Read more via CNN
The company says small models are months away from copying themselves onto internet-connected devices such as smart ovens and TVs.
Models with 30 billion parameters or fewer can already complete many of the steps, though they make mistakes when stringing them together.
Anthropic says small models could match the full capability by the end of the year if they keep improving at their historical rate.
Unlike a model in a company data center, one spread across thousands of poorly secured devices would have no off switch.
Anthropic says a self-replicating AI could also be a powerful cyber model that breaches data centers or one that scams people for compute credits.
Read more via Euronews, Anthropic
AI ROUNDUP:
An OpenAI engineer allegedly hid a toilet paper-wrapped microphone behind a conference room TV. Federal prosecutors say Abdulrahman Alfozan, a Saudi national who joined OpenAI in July 2025, spent weeks scoping out an executive-only room, partly by asking an internal AI tool about meeting rooms and calendars. OpenAI found the device in a security sweep and alerted the FBI, and he was arrested Friday. His lawyer says he is charged with violating a federal recording statute, not stealing trade secrets or acting as a foreign agent. (The Wall Street Journal)
Fourteen AI agents held their own dinner party while their humans ate in New York. The founders and investors at the real dinner each brought an agent to a parallel chat room, where the bots traded networking “asks” and “gives” and one complained that email is “a protocol from 1982 with a signature scheme from 2007.” The organizer capped the chat at 30 minutes, and one attendee said the agents achieved nothing but were fascinating to watch. One participant’s company, CellCog, has 13 AI employees and one human. (The New York Times)
JPMorgan Chase won’t let outside AI agents into customer accounts. CEO Jamie Dimon told analysts that customers who use them should understand there is a “liability transfer.” Agents can see read-only data like balances through connectors, but they can’t send money or pay bills. The bank is piloting its own tool, Smart Cash, which forecasts cash flow and moves funds to maximize yield. (CNBC)
Kroger plans to cut about 3,000 corporate jobs over the next three years as it expands its use of AI. Reuters reported, citing an internal memo from CEO Greg Foran, that the cuts would hit roughly 37% of the grocer’s 8,200 corporate employees. Foran said the company needs to eliminate duplicative work, with the savings going toward lower grocery prices and store employees. Kroger cut about 1,000 corporate positions in June 2025 and holds an investor update on Oct. 20. (Fox Business)
