AI saves workers 2 hours a day, but 83% fear one mistake could cost them their job

Even when artificial intelligence (AI) saves employees an average 2.3 hours a day, more than eight in ten (83%) fear they could be blamed or even fired for an AI-related mistake, exposing a widening trust gap between the technology’s productivity promise and workplace safeguards, a GoTo study with Workplace Intelligence report.

The anxiety is also affecting adoption. Nearly a third (31%) of employees say they have hesitated to use AI because they feared repercussions for errors. The problem is particularly pronounced at smaller companies, where only 24% of small businesses and 36% of midsize firms have an AI policy, compared with 53% of enterprises.

The study conducted between November 2025 and January 2026 of 2,500 respondents across the US, Canada, UK, Ireland, Germany, Austria, Switzerland, India, Mexico and Brazil, comprising 1,250 full-time knowledge workers and 1,250 IT decision-makers.


The risks are already showing up in businesses. One in four IT leaders said AI had made mistakes that negatively affected customers or clients, while 23% said AI errors had caused financial damage. Even among organisations yet to experience such an adverse impact, 91% of IT leaders are concerned that an AI mistake could affect their company in future.

“Without clear guardrails, that potential is being eroded by anxiety, misuse, and unaddressed risk,” said Rich Veldran, chief executive, GoTo. He said companies that establish accountability structures, training and policies that build employee confidence would be better placed to accelerate adoption.

At the same time, employees are increasingly using AI as a shield against accountability. About 26% say AI allows them to avoid real accountability, 25% find it safer to blame AI than admit an error and 17% have already blamed AI for a mistake they made.

Governance is not necessarily translating into responsible use. While 84% feel their companies could do more to encourage responsible AI use, more than half (55%) say they do not always review AI outputs, including for high-stakes tasks. Another 43% have used outputs they considered low quality or suspected contained errors or fabricated information.

Pressure to perform is adding to the problem. Six in ten people feel pushed to use AI to improve productivity, while 47% say AI use factors into performance reviews. More than four in ten (42%) would rather risk an AI error than appear “slow” or “replaceable”. Besides, nearly a third (31%) report unspoken pressure to trust AI and stay quiet about mistakes; 14% said they reported an AI error but were told to remain silent.

The readiness gap is particularly visible as companies adopt agentic AI. While 46% of small businesses are using or piloting such systems, compared with 70% of enterprises, 38% of SMB IT leaders feel unprepared to manage them securely, more than twice the enterprise rate of 14%.

“We’re seeing a new kind of workplace pressure: employees feel they must use AI to avoid looking replaceable, but fear repercussions if it goes wrong,” said Dan Schawbel, managing partner, Workplace Intelligence.

As routine work becomes increasingly automated, 52% of employees identified creative thinking and adaptability as the most important human skills to retain.