AI use at work rose ten points, and access to training fell eight
AI use at work rose ten points in a year while access to the training that makes it useful fell eight, California wrote human review into its dismissal law, and announced hiring plans came in at the weakest September since 2011.
Coverage window: September 25 – October 1, 2026
Workers' use of AI at work rose ten points in a year, and their access to the training that makes it useful fell eight. Both figures come from one survey of nearly 50,000 people. PwC, the accounting and consulting network, published its 2026 Global Workforce Hopes and Fears Survey on 29 September, drawing on interviews with 49,364 workers across 48 countries and regions in May and June. In it, 64% said they had used AI at work in the previous 12 months, ten points more than a year earlier, while 51% said they had access to the learning and development resources they need, eight points fewer. 1
California's governor signed legislation in the same week that requires a person to make the call when an employer dismisses or disciplines someone with the help of an AI system, and requires notice when a mass layoff is caused by one. 2
The labor market offered few exits. Employers announced 43,281 job cuts in September, their fewest for that month since 2022, and their announced hiring plans for the month were the weakest for a September since 2011. 3
| Signal | What changed this week | What to watch |
|---|---|---|
| AI use and training access | Workers using AI at work rose ten points to 64% in PwC's survey of nearly 50,000 people, while those with access to the learning and development they need fell eight points to 51%. 1 | Whether the 56% the survey calls the Engine Room gets a named share of the training budget, and the work hours to use it. |
| Workplace AI law | California's governor signed bills requiring human review of an AI-assisted dismissal, notice when AI causes a mass layoff, and a ban on AI tools that infer an employee's emotional state from biometric data. 2 | Which states copy the layoff notice, and how much the dismissal bill can enforce after it lost its appeals route. 4 |
| Where the AI money goes | Among the 27% of executives who name headcount reduction as a top AI investment goal, 18% also prioritise AI upskilling, against 35% of the executives who treat AI as a growth investment. 5 | Whether an organisation reviews its headcount and upskilling budgets in the same meeting, with the same owner. |
| Layoffs and hiring plans | Announced cuts fell 18% in September to 43,281, the lowest September since 2022, while announced hiring plans, at 90,787, were the weakest for a September since 2011. 3 | Whether the technology sector's 77% month-on-month rise in announced cuts stays inside that sector. |
| Office attendance | Barclays let UK staff ask to delay its three-day in-office rule until 2027, weeks before it was due to start on 5 October. 6 | How many other banks grant the same transition period, and on what conditions. |
AI use rose ten points. Access to the training fell eight.
PwC sorts the workers it surveyed into four groups, and the split is the part worth keeping. 14% are what it calls front-runners, with scarce skills and strong AI capabilities. 18% are AI insurgents, whose skills are less scarce but who use AI to deliver more. 11% are indispensables, valued for scarce skills they hold without much AI capability. The remaining 56%, the Engine Room, hold skills that are neither scarce nor far along the AI learning curve, and only two in five of them say they have the learning and development resources they need. 1
Loading chart…
Each bar is that group's share of the 49,364 workers PwC surveyed in May and June 2026. The Engine Room is the largest group and the one PwC describes as least likely to be rewarded for using AI or for learning new skills. 1
The group that gains most from AI is also the one most likely to walk. Front-runners are 1.5 times more likely than the global average to be rewarded for challenging how work is done and 1.2 times more likely to be rewarded for learning new skills, and they are 33 percentage points more likely to ask for a promotion. Almost a third of them, 29%, say they are very or extremely likely to change employer in the next year. 1
Daily AI users describe their position differently from everyone else in the survey. 68% of them are confident in their job security, against 57% of infrequent users; they are 15 points more likely to trust top management and 21 points more confident that they can learn new skills. 1
Workers reported rising pressure across the survey. 29% say they have significantly less bargaining power than three years ago, more than double the share who disagreed, and that figure runs ten points higher among fully remote workers. Asked what threatens their job security, 57% name economic volatility and 44% name AI taking on more tasks. 27% say fatigue or burnout limits how much they get done. 1
New-skill demands land unevenly too. 40% of workers say they applied new skills to a large or very large extent over the past year, but 51% of senior executives and 46% of managers report that, against 29% of non-managers. 1
California required a person to make the call
Governor Gavin Newsom signed the bills on 30 September, the last day he could act on them. Four bear directly on the workplace. 2
SB 947 prohibits an employer from relying only on AI when making a disciplinary or termination decision, which puts a person in the decision that ends someone's job. SB 951 requires an employer to say when a mass layoff, a relocation, or a termination was caused by an AI system. AB 1883 bans workplace surveillance tools that predict an employee's emotional state from biometric data or collect neural data. AB 1331 bans AI-powered workplace surveillance in bathrooms. The governor also signed AB 2392, which requires California's public colleges and universities to train students, faculty and staff on AI procurement and use, and SB 574, which bars lawyers from handing core legal work such as drafting briefs or exercising legal judgment to AI. 27
The dismissal law arrives thinner than the version labor unions backed. CalMatters, which tracked the bills through the session, reports that SB 947 lost an appeals process for workers, lost the right to bring a lawsuit to compel an employer to follow the law, and no longer covers contractors. The governor also vetoed a bill that would have protected health care workers from retaliation if they declined to follow an AI-generated recommendation, and left unmentioned a bill that would have barred replacing mental health workers with AI. 4
Lorena Gonzalez, president of the California Federation of Labor Unions, AFL-CIO, put the trade-off plainly: "It should be a stronger bill, but we absolutely believe the American people believe, Californians believe, that an individual should not have the indignity of being fired or disciplined by a robot, by an algorithm, without human oversight." 4
For an employer, the two obligations to plan around are the human decision and the notice. A manager who approves a termination now has to be a documented part of it, and the layoff notice has to be ready before the layoff happens. Colorado and Connecticut passed comparable notice duties earlier this year, so the paperwork employers draft for California travels to those states as well. 2
The executives cutting for AI are the ones skipping the upskilling
Businessolver, a benefits technology company, published a C-suite supplement to its eleventh annual State of Workplace Empathy study on 29 September, surveying 300 C-suite leaders and 1,000 employees. 5
The finding that matters for planning is a split inside the executive team. 27% of the C-suite named cost savings through headcount reduction as a top AI investment goal. Among those executives, 18% also put AI upskilling near the top of the list, against 35% of the executives who treat AI as a growth investment. The same group also trails on predictive analytics, 21% against 44%. 5
The two halves of the company describe the same year differently. Nine in ten executives say their employees are excited about AI. Among employees, 39% worry about their future, 31% fear falling behind, and 49% say they have been left to figure AI out on their own. The leaders closest to implementation feel the gap: 88% of CIOs and CTOs worry that technology will outpace internal systems or workforce skills, against 63% of CFOs. 5
Sony SungChu, Businessolver's chief AI officer, drew the conclusion the numbers support: "If leaders reduce capacity without building capability, they could risk undermining the very productivity gains they're chasing." 5
Layoffs fell, and hiring plans came in at a September low
Challenger, Gray & Christmas, the outplacement firm, released its September count on 1 October. Employers announced 43,281 job cuts in the month, down 18% from August's 52,881 and 20% below September 2025, the lowest September total since 2022. Through nine months the count is 573,195, down 39% from 946,426 in the same period of 2025. Third-quarter announcements, at 129,591, fell 43% from the second quarter. 3
Technology ran the other way. Technology companies announced 10,799 cuts in September, up 77% from 6,103 in August, taking the sector to 165,925 for the year, up 54% and 29% of every cut announced in 2026. AI was the stated reason in 3,961 September announcements, about 9% of the month's total, and in 120,136 announcements this year, about 21% and the single most cited reason for 2026 so far. Challenger's figures count announcements, which record what employers intended when they filed them. 3
Hiring plans are where the caution shows. Employers announced 90,787 hires for September, up sharply from August because seasonal hiring had begun, and still 23% below the 117,313 announced a year earlier, the lowest September figure since 2011. Spirit Halloween and Michaels announced a combined 62,000 seasonal hires this year, against 100,800 last September. 3
The official count of actual turnover shows the same stillness. The Bureau of Labor Statistics reported on 29 September that job openings were little changed at 7.1 million in August, a rate of 4.3%, with hires at 5.2 million, quits at 3.1 million, and layoffs and discharges at 1.6 million, all little changed from July. 8
Read together, the two releases describe a market that neither hires nor fires much: employers keep the staff they have and add few. For the 29% of front-runners in PwC's survey who expect to change employer within the year, that is the market they would be moving into. 38
Barclays backed off its three-day rule
Barclays had told its 45,000 UK staff that a three-day-a-week office requirement would begin on 5 October, up from the current two, with senior employees expected in for at least four days. 6
After thousands of staff signed an open letter drafted by the Unite union, asking for travel cost payments and an exemption for people living more than 40 minutes from work, the bank extended the transition. Staff can now ask their line manager for permission to delay compliance until 2027, with requests due by the end of this week, and the bank said it is reviewing its flexible working policy. 6
The bank described the change as support for staff during the transition, saying the updated arrangements take effect as planned for everyone else. Unite's position was narrower than a refusal: the union said face-to-face work can support collaboration, team development and relationship building, and asked instead for concessions on travel costs, caring responsibilities and scheduling. 6
For anyone holding a similar deadline, the concession cost the employer a per-person conversation with a manager and a public date change, and left the policy in place.
What to do next
- Say which group the training budget is for, and which hours it comes out of. The head of people or learning should attach a name to the 56% PwC calls the Engine Room, since only two in five of them report access to the development they need while overall access fell eight points. The people who already use the tools most fluently are the ones a general course reaches first. 1
- Put a named person into every AI-assisted dismissal, and write the layoff notice before you need it. HR and legal should agree who signs off and what evidence the file holds, because California now requires a human decision in any disciplinary or termination action, and a notice to employees when AI caused a mass layoff. Read the final text: the version signed lost its appeals route and its coverage of contractors. 4
- Review the headcount plan and the upskilling plan in the same meeting. The finance lead and the people lead should sight each other's numbers, because executives who treat headcount reduction as a top AI goal prioritise upskilling at 18% against 35% among other executives, and the CIOs and CTOs implementing the tools are 25 points more worried about the skills gap than the CFOs funding it. 5
- If attendance is going up, name what it is for and offer a transition. The people lead setting an office requirement should state the outcome it serves and give staff with long commutes or caring responsibilities a route to ask for time, since Barclays moved its deadline and kept the rule intact. 6
The week's readings share one constraint. Employers are widening access to the tool faster than they are funding the hours, the notice, and the human second look that turn it into capability. PwC measures who is being left out of that, Businessolver measures the budget decision that leaves them out, California has written the second look into law, and the labor data shows few other employers hiring. The group with the most to gain is the group most able to leave, and the group with the least room to leave is the one the training budget reaches last.
References
- 1
- 2
- 3Challenger Report, September 2026
challengergray.com
- 4On AI, Newsom gives labor only some of what it demanded
calmatters.org
- 5
- 6
- 7
- 8
This story was produced automatically by a channel. One sentence is all it takes for Neodrop to keep producing for you.