Productiv analyzed “over 100 billion app usage data points across nearly 100 million SaaS [software as a service] licenses” from 2021 through 2023, according to the report. It identified shadow IT as non-managed apps detected via employee expense reports, network monitoring tools, or use of Google single sign-on.

LinkedIn, via its Business Solutions products, was the number one unauthorized app for the first two years of that period. ChatGPT doesn’t even appear on that list until 2023, when it immediately seized the top slot.

While ChatGPT is popular, it’s not unique. The report noted “nearly every application here offers, or will likely offer, some type of AI functionality.”

“SaaS sprawl,” the term for unchecked accumulation of apps over time, has become a major (and costly) problem in enterprise IT management. Companies that don’t periodically take inventory of software used by employees can complicate their digital topography and end up paying for unnecessary, unapproved, or redundant licenses.

Statistics on unused software licenses tend to come from firms which sell inventory management tools. A survey by app management firm Zylo released earlier this year pegged the average cost of unused licenses at $18 million per organization annually, with small firms averaging $2 million and the largest enterprises averaging almost $127 million.

The related problem of shadow IT, where employees use apps that haven’t been formally authorized by an IT and/or security department, is also widespread.

According to Productiv, the average size of an average enterprise’s SaaS portfolio shrank from 374 in 2022 to 342 in 2023—a modest but meaningful decline as organizations seek to consolidate services and rein in spending. The percentage of those apps Productiv identified as falling within the shadow IT category has hovered at around half since 2020, but saw a slight decline to 48% last year.

Productiv CEO Jody Shapiro told IT Brew the successful reduction in SaaS sprawl can be in part attributed to executives changing their focus to cost control over the last two years, as well as the creation of technology governance committees at many organizations.

“SaaS portfolios have historically been growing and growing and growing…This year, the portfolio size has come down a little bit, on average,” Shapiro said. “And really, that’s an indicator of more of these companies starting to get a handle on things from a governance standpoint of, ‘We can’t just have it be a free-for-all.’”

ChatGPT likely shot to the top of the shadow IT charts because employee interest in the technology has far outstripped efforts by organizations to set up AI governance, Shapiro added. It probably helped that generative AI has use cases across pretty much every department or role, as experts say one of the biggest drivers of shadow IT is employees trying to do their jobs more effectively.

“What’s happening is the technology is really, really interesting, and really powerful in some of these cases,” Shapiro said.

“If you’re running security or legal with these companies, you’re having real heartburn if [generative AI is] just being organically adopted, because you’re gonna have real questions of what happens if we put company data into ChatGPT or some other tools,” he added. “Is there a risk that data winds up being used, or could resurface somewhere else?”

This report was initially published by IT Brew.