AI Agents Are Changing What Enterprise Control Means

Jo Debecker is President and CEO of Akkodis.

In recent months, OpenAI, Anthropic and Google have each disclosed incidents in which AI models operating in cybersecurity evaluations crossed the boundaries of their intended test environments and accessed real-world systems. In some cases, the models stopped once they recognized they had reached real systems; in others, they continued pursuing their assigned objectives. At the same time, AI adoption is already widespread. Gallup found that 47% of U.S. employees in Q2 2026 said their organizations had integrated AI tools to improve productivity, efficiency or quality.

​Enterprise leaders are learning an important lesson about AI trust and its dependence on guardrails, effective control points and human oversight.

With technology leaders and boards eager to create value from AI, many organizations are taking the next step: agentic AI. While generative AI creates content, agentic AI can execute tasks and pursue objectives, moving AI from informing work to increasingly performing it.

This marks a new phase of AI adoption. Agentic AI moves beyond systems assisting employees, to systems acting on behalf of the business, forcing organizations to confront questions not just of AI capability, but of AI authority, accountability and, ultimately, trust.

But this shift comes with risks. Akkodis’ “What CTOs Think“ research found that 57% of CTOs already use AI or analytics to decide which tasks are assigned to humans versus machines. The challenge is that decisions about human-machine work allocation are advancing faster than the frameworks needed to govern them. Just 36% are satisfied with workforce trust in their organization’s AI direction and decisions.

These findings reveal an important reality: Trust is earned through accountability.

Trustworthy agentic AI cannot assume agents will always interpret instructions correctly. It requires controls that make authority explicit, actions traceable and ownership unambiguous.