- Half of the IT teams do not fully know what their AI agents are accessing daily
- Businesses love AI agents, but also fear what they are doing behind closed digital gates
- AI tools now need governance, audit trails and controls like human employees
Despite the growing enthusiasm of the Agent AI in the business, new research suggests that the rapid expansion of these tools is pushing for them to secure them.
The cell -point survey of 353 IT professionals with enterprise security responsibilities revealed a complex compound of hope and anxiety on AI agents.
The survey reports that 98 % of organizations intend to increase the use of their AI agents within the coming years.
AI agents have left behind security preparations in adoption
AI agents are being integrated into operations that handle sensitive enterprise data, from consumer records and finance to legal documents and supply chain transactions – however, 96 % of respondents said they see these agents as a growing security threat.
A primary problem is visible: Only 54 % of professionals claim that their agents are fully aware of the data – which leaves almost half the enterprise environment in the dark in the dark on how AI agents interact with important information.
By promoting the issue, 92 % of the survey agreed that it is very important for security to govern AI agents, but only 44 % of the actual policy is available.
In addition, eight out of ten companies say their AI agents have taken steps that they did not intend – it includes access to unauthorized system (39 %), sharing inappropriate data (33 %), and downloading sensitive content (32 %).
Even more disturbing, 23 % of respondents admitted that their AI agents have been cheated for displaying access to a potential gold mine for malicious actors.
One notable insight is that 72 % believe that AI agents offer more risks than traditional machine identification.
One of the reasons is that AI agents often need numerous identities to work effectively, especially when connected with high -performance AI tools or systems used for development and writing.
Identification is getting louder for change in the first model, but cell points and others say that organizations need to treat human consumers such as AI agents, access control, accountability procedures, and full audit trails.
The AI ​​agent is a relatively new addition to the business space, and it will take time for organizations to completely integrate them into their work.
The Cell Point concluded, “Many organizations are still early in this journey, and the growing concerns about controlling the data highlight the need for strong, more comprehensive strategies to protect identity.”