Former Employees Keep AI Access, Putting Company Data at Risk

By Tom Whitfield

24% of former employees said they could still use a company AI tool or assistant after leaving, according to Sagiss’ 2026 Managed Security Report: Employee Offboarding, published in October. The managed IT and cybersecurity provider also found that 16% said company information remained in an AI tool or work chat history.

That information could stay at risk and prove harder to recover. Pollfish surveyed 1,000 U.S. adults in September 2026 who had left jobs involving company technology within the previous five years. These findings are self-reported, not verified audits of corporate accounts, but they expose a gap between a completed offboarding ticket and verifiable access termination.

The account closed. The data did not.

Travis Springer, Sagiss president, said the survey could not determine whether former employees still accessed company-managed assistants, personal AI accounts used for work, or both. Some cases may involve company-managed assistants that were not disabled; others may involve personal accounts used for work. The survey could not establish which situation was more common, and each requires a different response.

A company can disable an account it manages. Information copied into a personal AI service may remain available through an account the company does not control.

“Disabling an account is only part of the job. You also have to think about what company information has already been put into these tools and who controls it,” Springer said.

The residual risk can be harder to reverse because the organization must first establish what left with the employee. A green offboarding ticket is not, by itself, evidence of termination.

The perimeter extends beyond chat tools

Security consultant Dave Shackleford said AI-enabled SaaS applications can query business systems, modify records and invoke APIs on a user’s behalf. He also identified three security blind spots:

• Service accounts
• API tokens
• OAuth grants

Some can be long-lived or poorly tracked. AI agents can inherit employee permissions, expanding access through those connections. Organizations therefore need an inventory of embedded AI capabilities, connected agents and nonhuman identities, not merely the applications employees use.

HR knows. The manager knows. IT may not.

Springer said Sagiss sometimes finds accounts belonging to departed employees during routine reviews because IT was never told they had left.

“We regularly review clients’ inactive accounts and identify people who left the company without IT ever being notified,” he said.

The survey found that 33% of respondents said responsibility for terminating access was unclear or that nobody appeared to handle it.

“HR knows the employee is leaving, the manager knows which tools that person used, and IT manages the accounts it knows about. Sometimes those pieces never come together,” Springer said.

Springer thinks HR should see the offboarding process through, even when IT and security shut accounts and verify access. Companies may divide those duties differently. Failure occurs when no one checks what was missed.

Finding the tools workers actually use

At Flexera, CIO and CISO Conal Gallagher surveyed employees to identify informal AI use, including tools introduced without formal IT approval, and determine which applications helped them work. The exercise surfaced software adopted independently by business teams.

“Our first step was to bring informal AI use cases into the open by engaging employees,” Gallagher told TechTarget.

At cryptocurrency payments company MoonPay, CIO and CISO Doug Innocenti used AI security software to examine applications and their underlying models, including how providers might use company data.

“We were able to perform an independent review on tools and understand which ones we wanted to pick,” Innocenti told TechTarget’s Beth Pariseau.

Both companies addressed the same prerequisite for reliable offboarding: knowing which applications employees use and where company information may flow.

One surviving account changes the risk

Among survey respondents who had used seven or more work accounts, 87% said at least one still worked after they left. Even among those who had used one to three accounts, 71% reported the same problem.

Single sign-on simplifies revocation by letting IT disable one central account rather than locate every application login. It does not automatically remove access already granted. Microsoft notes that shutting off an employee’s access may leave active sessions running; some applications continue accepting previously issued tokens or maintain their own sessions.

What does this change for offboarding?

It moves the finish line from disabling a central login to proving that access and data exposure have ended. Springer recommends checking applications outside SSO, shared credentials and surviving sessions. He also wants CIOs to determine what company information employees stored in AI services and whether administrators still control it.

Organizations should record evidence of the shutdown:

• Which accounts were disabled
• When each account was shut down
• Who verified the change

They should then investigate delays, omitted applications and unexpected login activity after departure. Some data may remain beyond the organization’s enforcement reach, a limit worth documenting rather than discovering later.

“You should be able to confirm someone is off the systems instead of finding out weeks later that an account still works,” Springer said.