Organizations must treat every agent ‘as a persistent digital actor without conscience or incentive’
HR departments could be exposed to liability, privacy, and workplace-fairness risks from autonomous AI agents already operating inside their organizations without clear ownership or oversight, according to new research.
The Govern Enterprise AI Agents While Preserving Innovation report by the Info-Tech Research Group states that AI agents cannot be governed like traditional software or human staff because they act autonomously across systems.
"AI agents cannot be governed like traditional IT assets or earlier AI models because they do more than generate outputs; they act across systems," says Altaz Valani, principal advisory director at the Info-Tech Research Group, in a press release.
The firm identified five recurring gaps:
- shadow AI built outside sanctioned tools
- mismatched autonomy and validation
- "runtime drift" as agent scope expands unnoticed
- unmanaged access permissions
- and ambiguous ownership when an agent causes harm
Most employers appear ready to embrace agentic AI – the next evolution of AI, according to a previous KPMG report.
Adoption outpacing oversight
The Info-Tech Research Group reports that 88% of organizations now use AI in at least one business function, but two-thirds remain in pilot mode rather than scaling – a gap the firm attributes to weak governance maturity.
Citing a 2026 Informatica survey, the research found 47% of organizations have adopted agentic AI, up from near-zero two years earlier, while 76% of those leaders said governance has not kept pace with employee use.
The firm also cites 2025 McKinsey & Company research showing 51% of AI-using organizations have experienced at least one negative consequence, most often inaccuracy. Also, Strata Identity 2026 data show that 15% of employees run unauthorized "shadow agents" — activity HR may have no visibility into when it touches employee-facing systems.
Firm recommends phased response
The Info-Tech Research Group recommended a three-phase approach:
- establishing governance authority and guardrails
- mapping the agent lifecycle and classifying agents by risk,
- and operationalizing accountability through defined ownership and executive reporting
Manish Jain, principal research director at Info-Tech, says in the report that "the right agentic AI governance model starts with treating every agent as a persistent digital actor without conscience or incentive," adding that governance "must enable innovation" rather than rely solely on upfront approvals.
The firm pointed to a case reported by Palo Alto Networks involving 3CX, in which a low-code automation agent with persistent credentials and no clear human owner caused cascading failures once a downstream system changed, as an example of the accountability gaps organizations face.
HRD previously reported that a growing number of companies have stopped calling their AI systems tools. They give them names, job titles and a spot on the org chart, and according to a research, that shift comes with a cost nobody planned for: The moment managers start thinking of an AI as a coworker rather than software, they start reviewing its work less carefully.
Governed property, companies can gain a lot with the use of agentic AI, according to previous reports:
|
Benefit |
Supporting data |
Source |
|
Higher individual and team productivity |
66% of organizations adopting AI agents report measurable productivity gains, and separately, 80% of respondents say AI has improved their individual productivity |
PwC, AI Agent Survey (2025), McKinsey & Company, The State of AI in 2026: On the Road to ROI (Aug. 25, 2026) |
|
Reduced operating costs |
57% of organizations adopting AI agents report realizing cost savings |
PwC, AI Agent Survey (2025) |
|
Faster, better-informed decision-making |
55% of organizations report faster decision-making from AI agents, and 50% of individual respondents say AI helps them make better decisions |
PwC, AI Agent Survey (2025), McKinsey & Company, The State of AI in 2026 (Aug. 25, 2026) |
|
Improved customer experience |
54% of organizations adopting AI agents report improved customer experience as a direct benefit |
PwC, AI Agent Survey (2025) |
|
Lower software and vendor spend through in-house building |
32% of organizations decided against purchasing at least one software product or feature because agentic coding tools let them build the functionality internally; scaling of agents at large enterprises rose to 40%, up from 27% a year earlier |
McKinsey & Company, The State of AI in 2026: On the Road to ROI (Aug. 25, 2026) |
