Crypto Briefing • October 9th 2026, 6:40 PM
Enterprise trust in autonomous AI agents slips, VentureBeat survey finds
Key Summary
A recent VentureBeat survey found that the share of companies allowing or pursuing automated production changes has dropped from 75% in July to 56% in August, indicating a decline in enterprise trust in autonomous AI agents. The survey also revealed that 61% of organizations that run pre-deployment evaluations reported at least one case where an AI feature passed testing and then failed in a way that affected customers.
Please see our real time news feed on our Home Page
Enterprise Trust in Autonomous AI Agents Slips Amidst Concerns Over Evaluations
What the Numbers Show
The August data comes from 140 respondents. The core question was whether organizations let AI agents make production changes based solely on automated evaluations, with no human approving the final step.What the Numbers Show (continued)
The respondent pool also shifted. Decision-makers made up 53% of participants in August, up from 44% in July.The Problem with Passing the Test
The most uncomfortable number in the survey concerns evaluations themselves. Among organizations that run pre-deployment evaluations, 61% reported at least one case in the past 12 months where an AI feature passed testing and then failed in a way that affected customers.What This Means
It is worth keeping the scale in perspective. The August survey covers 140 respondents, and the composition of the pool changed between months, with more decision-makers in the latest round.What to Watch Next
It is worth tracking whether the September and October figures hold near 56% or bounce back toward July's level. Also worth tracking: whether the 20% engineering toward autonomous low-risk deployments actually gets there within their 12-month window, or quietly joins the 35% that have opted out.#AI#US#Crypto#Tech