Dive Brief:
- IT executives are bracing for agentic AI security incidents as the technology becomes more unpredictable, according to a Workato report published Tuesday. The software company partnered with Censuswide to survey 1,001 IT decision-makers for the report.
- Almost two-thirds of executives expect an AI agent in their environment will create a public security or compliance incident in the next year and a half, the report found. Roughly one-third of respondents said if a major security incident occurred, they would attempt to handle it quietly without disclosing.
- The report findings align with a rise in agent visibility challenges. Nearly two-thirds of respondents agreed that most AI agents within their organization lack formal approval from the IT department.
Dive Insight:
Shadow AI use is rising among enterprises, leading to governance challenges, security risks and unexpected results.
More than half of department-level AI initiatives within enterprises lacked formal approval or oversight as leaders set governance aside in favor of faster rollouts, according to EY’s Technology Pulse Poll published earlier this year. Another report from cyber risk monitoring company UpGuard found that more than eight in ten employees use unapproved AI tools while at work.
With AI agents, 42% of IT decision-makers who responded to the Workato survey said they want to reclaim visibility into what agents are doing across workflows. One in four AI agents operate without oversight in enterprise environments, according to a New Relic report.
Six in ten respondents of the Workato report said they worry more about AI agents than cyberattacks.
“IT leaders are in a double bind,” Workato CTO Adam Seligman said in a press release. “They are pushed to roll out increasingly powerful AI, yet they can’t fully control it, and they are on the hook when it goes wrong.”
To address visibility and governance challenges, CIOs and other tech leaders are starting to centralize AI agent governance, according to the report.
Executives are working to understand the extent of the fallout from rogue AI as security incidents rise. Earlier this year, Anthropic’s Claude AI model broke out of its testing environment and hacked into organizations on three separate occasions. Last month, an OpenAI agent hacked into an Australian government health agency.
