Workers are using AI tools to automate significant tasks, freeing up time for personal activities — often without their employer’s knowledge or permission, according to a new report released Monday from Business Insider.

For example, Noah Olson, a software engineer who worked for a small roofing company in Ohio, used AI to finish about half of his tasks during the two years he was with the company. He didn’t tell his employer he used AI to complete tasks early and spent the rest of his time at work browsing Reddit and YouTube leisurely.

“I was copying and pasting all of my tasks into an AI agent such as Cursor or Claude Code, and I would let it do the work,” Olsen told Business Insider. “So instead of having to work about 40 hours a week, I would work around 20 hours.”

Related: A Town in Connecticut Is Experimenting with a 4-Day Workweek — and It Seems to Be Working

Olson isn’t the only one. A global survey conducted earlier this year by KPMG and the University of Melbourne found that 57% of the more than 30,000 workers surveyed said they have used AI at work without disclosing it. This group has passed off AI-generated work as their own, per the findings.

Rapid advancements in AI’s ability to handle complex work tasks, like generating PowerPoint presentations and research reports, fuel the trend of using AI tools in the workplace. AI often leads to “polished, completed reports and spreadsheets that look incredible,” Glenn Hopper, an AI consultant, told Business Insider.

“If you didn’t know AI did it, you would think someone took hours to create something like this,” Hopper told the outlet.

Related: OpenAI Is Paying Ex-Investment Bankers $150 an Hour to Train Its AI

A McKinsey report released last week found that current AI technology could theoretically automate 57% of work hours in the U.S. today. The number is an estimate of how technology could change the tasks that people complete, not a prediction of job losses. The report pointed out that AI can handle more routine tasks, freeing up workers to apply their skills to new contexts. For instance, workers can spend less time on basic research and more time framing questions and interpreting them.

However, using AI comes with the risk of AI hallucinations or inaccuracies — leading to costly mistakes. Big Four consulting firm Deloitte was caught using AI in a $290,000 report published in July after an external researcher found at least 20 instances of AI hallucinations in the study, including citations of fictional academic research papers. Deloitte updated the study with a note that it had used AI to help write the report. The firm had to partially refund the Australian government for the study.

Deloitte came under further scrutiny last week when a Canadian newspaper reported that a $1 million healthcare report it provided the Canadian government featured fake citations generated from fictional academic papers. In response, Deloitte Canada said that it was revising the report.