AI agents are expected to transform enterprise cybersecurity, with 100% of UK IT leaders identifying at least one security concern that agents could improve, according to the latest research from Salesforce.

The CRM software provider’s State of IT: Security report found that almost seven in 10 (69%) of security leaders now recognise their security practices need to change in line with evolving threats, and see the implementation of AI agents as a new frontier in threat detection and real-time monitoring. 

Despite their optimism that AI will give them the tools they need, many security leaders see significant implementation roadblocks ahead‌.

According to Salesforce’s study, based on a global survey of over 2,000 enterprise IT security leaders, including 100 in the UK, more than half (54%) aren’t fully confident they have the right guardrails in place to deploy AI agents, while 49% worry their data foundation isn’t set up to get the most out of agentic AI.

That isn’t a new problem. Last October, a report from MIT Technology Review Insights found that poor data foundation is holding back the AI aspirations of more than three-quarters of organisations.

Meanwhile, research from SoftServe published just last month revealed a growing gap between companies’ AI ambitions and their data capabilities, with most firms struggling to manage or access quality data, despite pouring resources into AI tools. 

“Trusted AI agents are built on trusted data,” said Alice Steinglass, EVP & GM for Salesforce platform, integration and automation.

“IT security teams that prioritise data governance will be able to augment their security capabilities with agents while protecting data and staying compliant.” 

Adding to the growing data problem, Salesforce’s study found that less than half (47%) of IT leaders think they can deploy agentic AI with the right permissions and access protocols, a real concern, seeing as 54% of UK firms are not sure their employees know how to responsibly manage data around AI agents.

Recommended reading

• Salesforce: Business Leaders Are Losing Trust in Their Data
• UK Gov Using Personal Data to Develop ‘Murder Prediction’ Tech
• Smaller AI Models May Outpace LLMs, Gartner Predicts

Compliance issues are adding to the strain, too, with 80% of UK security leaders agreeing AI agents present serious compliance challenges, but just 37% are confident they can deploy this technology in line with current regulations and standards. 

Only 41% of security teams said they were ready for regulatory changes around AI, while the majority (69%) of firms are still struggling to wade through the complex maze of new standards and rules that they say are making AI compliance more difficult.

However, none of those issues seems to be seriously slowing down plans around agentic AI adoption.

While 41% of UK IT security teams already use agents in their day-to-day operations, Salesforce said that this number is expected to nearly double, with 77% of firms saying they will begin to use AI agents within the next two years.