Gravitee finds 2.4 million AI agents running with no human oversight

New research from Gravitee says one-third of AI agents deployed at major U.S. and U.K. firms are fully autonomous, with no human in the loop. The findings highlight a growing governance gap as companies deploy agents faster than they assign accountability or security controls.

Why it matters: - Gravitee says millions of AI agents are already making decisions and taking actions inside major firms without direct human oversight. - The research points to a governance gap that could affect security, spending, data handling and accountability. - Nearly a third of firms are deploying AI agents specifically to reduce headcount, raising labor-market implications as unemployment rises in the U.K. and employment rates stay flat in the U.S.

What happened: - Gravitee commissioned research that polled 750 CTOs and vice presidents of engineering across the U.S. and U.K. - The survey found 7,239,650 AI agents deployed in U.S. and U.K. businesses. - Of those, 2,445,553 are fully autonomous and operate with no human in the loop. - Gravitee said the number of AI agents appears to be doubling every six months. - Gravitee released the data ahead of the premiere of its documentary, “AI Agents: How Do We Take Back Control?”

The details: - Just 7% of organizations have named a single accountable person for AI agents. - Among organizations that do assign accountability, there is no clear consensus on whether the CIO, CDO or CPO should own the role. - Eight in 10 organizations said they feel pressure to deploy AI agents even when security is not fully in place. - For more than a third of organizations, that pressure comes from the boardroom. - Nearly a third of technology leaders said they are personally uncomfortable with the pace at which their organization is moving. - Gravitee CEO Rory Blundell warned that AI agents could cause data leaks, unwanted spending and deleted files without oversight. - The documentary is hosted by Alex Kantrowitz, founder of Big Technology. - Featured voices include Theresa Payton, Ramesh Raskar, Sharon Gai, Ambica Rajagopal and Rory Blundell. - More information is available at Gravitee AI Agent Documentary.

Between the lines: - The findings suggest companies are scaling agentic AI faster than they are defining ownership, governance and controls. - The board-level pressure reported by respondents suggests AI deployment is being treated as a competitive imperative, not just an IT initiative. - The lack of agreement on who should own AI-agent behavior signals an emerging management problem for enterprise leadership.

What's next: - Gravitee is using the documentary launch to push firms to “take back control” of their AI agents. - The research is likely to intensify debate over who should be accountable when autonomous systems act without approval. - Companies deploying AI agents may face growing pressure to establish named owners, guardrails and security checks before expansion continues.

The bottom line: - AI agents are spreading fast inside major firms, but accountability is not keeping pace.