A new global survey of human-resources leaders highlights how much generative AI has permeated work behind the scenes. In a poll released on 14 October by HR platform Oyster, 95% of companies reported that their employees are using AI tools in roles where the job description never included AI ([1]). Two-thirds of these companies have added AI skill requirements to traditionally non-technical jobs like marketing, HR and operations ([2]), often informally. In many cases, workers pick up generative AI on their own initiative without any change in title or pay.
The findings echo a UK study by Deloitte, which found 63% of British workers use GenAI tools in their jobs ([3]). Nearly one in ten have even used AI that their employer forbid ([4]), underscoring a growing “shadow AI” trend. Deloitte’s survey of 25,000 employees also found that 63% worry their managers might think AI can do their job ([5]). The research suggests employees see value in using AI for productivity, but also fear that doing so could make them redundant. For leaders, it’s a sign that staff are already quietly reshaping their own roles with AI, and that organisations risk losing control of how work is changing if they don’t guide these efforts.
Early evidence is emerging that AI-driven efficiency gains may come with a hidden cost in workforce growth. In the US Federal Reserve’s latest Beige Book survey of businesses, five out of twelve regions reported companies reducing or freezing hiring after boosting productivity with AI ([1]). One regional report noted that while “artificial intelligence was not causing layoffs, it was dampening the creation of new jobs” ([2]). Similarly, S&P Global’s international PMI survey showed a net employment impact of, 5 percentage points over the past year from AI adoption, with more firms cutting jobs than adding due to the technology ([3]). It even forecast a further net decline in headcount in the year ahead ([4]).
These cautionary signals contrast with rosier forecasts that AI will simply augment workers. Instead, they indicate that some firms are using efficiency gains from AI to trim future headcount growth. That doesn’t necessarily mean mass layoffs, several businesses told the Fed they were reducing staff through attrition or not backfilling roles, rather than firing employees ([5]). However, the effect could still be significant over time. Leaders should consider how to balance productivity improvements from AI with strategies for reskilling and evolving roles, to avoid stalling talent pipelines and morale.
What does a fully AI-integrated workplace look like? A new report from Anthropic, one of the leading AI development companies, offers a glimpse. The firm has “molded its workflows around AI” ([1]) by using its Claude chatbot as a coding assistant and general-purpose aide throughout the organisation. As a result, the composition of its workforce has shifted: the share of junior-level employees in technical roles has fallen from 21% to 14% since early 2025, and junior technical hires plunged from 27% to just 9% of new recruits ([2]). The company attributes this partly to AI systems taking over routine tasks, as well as a natural broadening of non-technical teams as the business scales ([3]). In fact, by August this year, Anthropic’s internal metrics show its AI agents were performing six times more work hours than human employees ([4]).
The report suggests that even as AI boosts output, it changes what humans do. Workers report spending more time overseeing AI and coordinating projects, an April survey found one-fifth of Anthropic’s staff said their jobs had shifted “from actually doing work to managing Claudes” ([5]). Employees described a constant push to "use Claude for everything," with one engineer likening the role to babysitting a swarm of AI assistants ([6]). Moreover, the flood of AI-generated results means human judgment and attention have become a “bottleneck” in productivity ([7]) ([8]). For executives, the Anthropic case illustrates both the potential and the pitfalls of deep AI adoption: greater output per employee, but also the need for new skill sets, oversight mechanisms and career paths as entry-level grunt work is increasingly automated.
Traditional industries are also transforming their workforces to compete in the AI era. On 18 October, Disney reached a deal to hire about 60 employees from Character.AI, a Silicon Valley chatbot startup that filed for bankruptcy ([1]). This represents the majority of Character.AI’s staff, who will receive roughly $60 million in total annual compensation from Disney ([2]). The move comes after Character.AI’s co-founder, Karandeep Anand, was recently appointed as Disney’s first-ever Chief Technology Officer ([3]). By absorbing most of the defunct AI company’s personnel, the 100-year-old entertainment giant is effectively buying an experienced AI team to accelerate its own transformation.
Disney’s aggressive push into AI talent underscores the growing demand for expertise that can embed advanced AI into legacy businesses. It also highlights a broader trend: where new AI startups falter, established companies are ready to snap up their talent and technology to innovate from within. For leaders, this signals intensifying competition for skilled AI professionals, and the value of integrating AI capabilities into core business functions, even for companies far outside the tech sector.