Skip to content
Theme
OnDevice Tools

GadgetsLaptops & PCs

Workers are sharing passwords with AI tools, Dashlane report finds

Nearly one in five UK employees has handed a password, API key or other login to an AI tool, according to research covered by TechRadar. Dashlane says 72% of UK security executives have already seen an incident involving an AI agent's access.

Illustration of an office worker pasting a password into an AI chatbot on a laptop

Key takeaways

  • Almost one in five UK employees has shared a password, API key or other login with an AI tool, Dashlane found.
  • 48% of surveyed workers let AI complete at least one task without prior approval, and one in ten allow AI to handle payments.
  • 72% of UK security executives reported an incident involving an AI agent's login or access in the previous 12 months.
  • 42% of workers say they have never been told which information is safe to share with AI.
In this article
  1. Workers hand logins to AI
  2. What the survey found about daily AI use
  3. Trust builds fast and access lingers
  4. Security teams report AI incidents
  5. The guidance gap employers need to close

Workers hand logins to AI

UK office workers are increasingly comfortable with sharing passwords with AI tools, according to research covered by TechRadar on 19 November 2026. The study, from Dashlane, found that almost one in five UK employees has passed an AI tool a password, an API key or another form of login.

An API key is a code that lets one piece of software communicate with another. Handing one over, along with a password, can give an AI tool a route into company email, documents and internal systems. Dashlane warns that the time saved can leave the employer exposed.

What the survey found about daily AI use

Illustration: AI, Dashlane logo, building, key and the labels AI Agents, Credential Theft, Dashlane

Approval is often missing. Dashlane says 48% of the workers it surveyed had let AI complete at least one task without checking with a manager first, including email, company documents and customer records. One in ten let AI handle payments.

The habit appears to be spreading faster than company policy. According to the report, 44% of workers had put AI to work in the previous year without getting approval first. Of those, 46% began before any workplace rules existed.

Security awareness is thin. Only a quarter said they always weigh up the risks before using an AI agent, a term for software that can carry out tasks on a person's behalf. Almost one in four said they had pasted confidential or commercially sensitive material into an AI tool that they would not share with a colleague. Dashlane also found that 42% say nobody has told them which data is safe to hand to an AI system.

Trust builds fast and access lingers

Workers appear willing to go further once a tool feels familiar. Dashlane reports that 37% said a single week of use would be enough for them to hand their login details to a newly adopted AI tool.

A further 45% said they would re-enter their login details each day if an AI tool asked them to. Dashlane links that habit to two risks: credential theft, where an attacker steals login details, and impersonation, where someone acts as if they were the account holder.

Access is also difficult to withdraw. The report says 15% had opened a system, file or account to an AI tool and later learned it should not have had that level of access. Of those, 43% think the access remains live.

Security teams report AI incidents

Security leaders describe a tidier picture than their colleagues do. Among UK IT security executives, 92% believe they have a full inventory of the AI agents running in their organisation. Yet a quarter of staff admit to using tools their employer has no knowledge of.

Incidents are already happening. Dashlane says 72% of UK security executives reported that a security incident had involved an AI agent's login or access in the previous 12 months. A third said an agent had reached data, systems or credentials it was not cleared to use, and 32% said an agent had taken an action it was not permitted to take.

Release speed is part of the strain. More than two-thirds of the executives, 68%, said they feel squeezed between faster AI adoption and keeping their organisation secure. Among that group, 54% said rollouts went ahead with less security review.

The guidance gap employers need to close

Dashlane's chief product officer, Christophe Frenet, said "The risks of AI spread faster". He said they turn up inside tools that have already been approved, and often develop without any further action from staff. He added that the company's data shows workers placing more trust in their new AI colleagues than in the people they work beside.

The findings point to a gap in communication rather than one faulty product. With 42% of workers reporting no guidance at all, many have no basis for deciding which tasks an AI tool may run on its own.

Staff who are unsure what they may share should ask their IT or security team before granting access, and follow any official guidance issued by their employer or software vendor. Dashlane's figures are self-reported survey results, so they show what workers say they do rather than independently verified behaviour.

Sources

  1. 'The risks of AI spread faster': Report finds workers are handing over logins and even passwords to AI to make their jobs easierTechRadar

Tags

Spotted a mistake? See our corrections policy. Read how News by OnDevice works in our editorial policy.