Stop Reviewing Your AI Agents. Start Governing Them.
Security Boulevard, Thursday, October 22nd, 2026
Checkpoint reviews can't keep up with AI agents; Wallarm argues for continuous governance with defined permissions and runtime monitoring.
In this vendor-authored response to a Brookings and World Economic Forum analysis, Wallarm's Tim Erlin argues the key questions for AI agents are what an agent did, whether it was authorized and who is accountable.
Stage-based reviews fail because agents invoke tools and act before humans look, so governance should be continuous, with permissions defined in advance, verifiable evidence and runtime monitoring.
He cites a Gartner prediction of more than 150,000 agents at the average Fortune 500 firm by 2028, with only 13% of organizations confident in their governance, and notes EU high-risk AI rules are delayed to 2027-2028.
He proposes Discover, Observe, Enforce and Govern steps.