AI enterprise deployment has governance, security concerns

They’ve given you a number, and taken away your name—but as the Johnny Rivers song goes, agents need to beware when taking chances.

That’s also true in the tech space, where enterprises are increasingly adopting agentic AI despite the risks involved. Peter Mullen, chief marketing officer at Kore.ai, told IT Brew that his company’s recent Agent Productivity Index found AI deployment is leading to missed connections and a rise in rework and IT challenges.

“Enterprises are deploying AI, whether it’s models, whether it’s agentic AI, whether it’s solutions at the top of the food chain or the bottom of the food chain—and it is causing challenges, it is causing people to redo their work,” Mullen said. “All of this is burning efficiency, it’s costing money, and it’s making people, in some cases, rethink how they want to go about doing things.”

Same old story. Issues around AI agents, governance, and security have plagued the tech industry for the past year. AI-related security incidents are a chronic problem, leading to data leakage, insecurity, and IT team rework. Industry reports regularly reveal governance challenges that IT teams must overcome, including cross-channel inconsistencies that can lead to miscommunication.

“Let’s say you’re chatting to the web chatbot, and you share relevant context about the situation you’re dealing with right now—maybe it’s a shoe return or something like that—then you get switched to a human agent,” Sophie Cheng, chief marketing officer at Sinch, told IT Brew in June. “If you don’t carry that context over, then you’ve got to repeat yourself.”

Numbers game. Kore surveyed over 400 IT leaders who are all actively running AI agents, 73% of whom are the final decision-makers on the organization’s stack. The numbers reveal that the stack is not as secure as one might think, and that risk is omnipresent due to how AI is being governed.

Majorities of the enterprises reviewed in the report reported problems throughout the process: 72% said AI agents are operating with unmanaged risk, 82% said agents are executing consequential actions without proper human oversight, 70% believed agents lack proper visibility and could not identify failures, and 93% said that reversing improper actions taken by agents ended up costing significant money.

Futurism. Addressing the problem is going to take a major reset in how enterprise businesses are run, Mullen told IT Brew. Governance and risk can’t be an afterthought, Mullen said, and survey results bear that out: As a result of agentic failure, 42% of respondents saw revenue loss from security incidents, and 31% from service-level agreements and agent malfunctions, while 28% experienced customer churn.

“It needs to start with the concept that this matters more than many of today’s current deliverables, and when you take that orientation, when businesses have the guts to take that perspective, it really changes how you think about things,” Mullen said.