According to a new, AI agents are being rapidly deployed within organizations Report From data governance firm Cellpoint.
Based on a global survey of over 350 IT professionals, the report found that the widespread hugs of agents – AI systems that are capable of preparing the plan and are taking action without human monitoring – is happening within a security vacuum. IT professionals replied, 84% said that their organizations already use agents internal, but more than half of the number (44%) have policies to control the behavior of agents currently.
Too: AI can erase half of the entry level white collar jobs in 5 years, CEO warns
Even more striking, 96% of respondents stated that they see agents as a security risk, yet 98% also said that their employers plan to expand the use of agents in the coming year.
Rise of agents
Agent Generative AI is the latest wave in innovation floods around AI, which began in Bayana after the release of OpenaiI chat in late 2022. Unlike traditional chatbots, which requires careful of human promises, agents have a lot of autonomy to perform functions.
(Disclosure: ZDNET’s original company Ziff Davis filed a case of April 2025 against Openai, alleging that it violates Ziff Davis copyright training and operating its AI system.)
Big tech companies are building and releasing agents in the draw in recent months, hoping to capitalize on generic AI Gold Rush. Small business, meanwhile-The new report of the cellpoint suggests-these are quick to embrace new systems, drawn by the charm of increased efficiency and sometimes by mounting pressure which is commonly described by mounting pressure-as mostly by large technology companies-one as fundamental disruptive technology.
Increased security risk
According to the report, the action of agents to take action without human oversight also makes them an increased cyber security risk.
“These autonomous agents are changing how the work is done, but they also offer a new attack surface.” “They often work with widespread access to sensitive systems and data, yet limited inspection.”
For the matter of enlightenment, the survey of the cellpoint revealed a dangerous irony: The vast majority of the respondents (92%) feel that the Adequate governance of AI agents is important for the safety of cyber security of its organizations, and yet many (80%) have also reported that agents have already worked in unexpected and potentially risky ways.
Also: AI agent adoption is increasing opportunities, dangers and IT budget.
Given their privileged and large -scale unexpected access to sensitive organizational data, Gyanasambandam recommends treating AI agents with the same security protocol that applies to human employees.
“As organizations have expanded their use of AI agents, they should take an identity-first approach to ensure that these agents are strictly controlled as human users, with real-time permission, at least privileges and full visibility in their actions,” they write.
Want more stories about AI? Sign up for innovationOur weekly newspapers.
