DXC Report: Companies Are Deploying AI Agents Faster Than They Can Secure Them

Companies around the world are speeding up their adoption of artificial intelligence, but security governance is not keeping pace. That is the conclusion of a new DXC Technology report, "Vision to Value: Closing the AI Execution Gap," published on September 4, 2026.

The gap between ambition and reality is stark. While nearly eight in ten business leaders rank artificial intelligence at the top of their priority list, fewer than half of organizations can claim mature processes that allow them to safely run AI agents operating autonomously inside corporate infrastructure.

Where the gap comes from

The report's authors point out that traditional security mechanisms were designed around users, devices, and networks as access control points. AI agents break that model because they act autonomously, carry out tasks across multiple systems at once, and require data access that is difficult to constrain using classic methods.

The problem mainly concerns control over which corporate resources an AI agent actually reaches, and whether its actions can be traced after the fact. In practice, this means security teams often don't know what data a deployed AI system is processing until an incident occurs.

As enterprises move from AI pilots to production use, security and governance must evolve just as fast - Halina Frańczak, Market Leader, DXC Technology Poland

The market's answer: Zero Trust for agents

In response to these gaps, DXC Technology is highlighting a partnership with Primary, a company building a Zero Trust platform designed specifically to oversee AI agents. The solution is meant to monitor agents' access to corporate data, manage their digital identities, and ensure compliance with regulatory requirements.

DXC states that it is the exclusive managed services partner for the platform, meaning the company intends to build a separate business line around securing AI agents rather than simply advising clients on the topic alongside other deployments.

What it means for companies in Poland

For Polish businesses, the report's findings carry weight, since domestic companies show a pattern similar to the global market: rising AI spending paired with a lack of a coherent risk management strategy. Earlier analyses showed that Polish companies are spending millions of zlotys on artificial intelligence without a clearly defined strategy, while regulators still lack fully formed regulatory frameworks for autonomously operating AI systems.

The growing autonomy of AI agents, capable of logging into systems on their own, searching databases, or initiating transactions, means the classic approach based on passwords and user permissions is no longer enough. Companies that fail to build an oversight layer around what their agents do risk data leaks or unauthorized access to internal systems.

The DXC report fits into a broader trend of industry warnings about the security of AI deployments in business. More and more enterprise vendors are developing dedicated tools to monitor and restrict agent permissions, treating it as a distinct cybersecurity market segment rather than an add-on to existing services.