# Keeping a human in the loop is one part of AI agent resilience

DevFeed: [Keeping a human in the loop is one part of AI agent resilience](<https://devfeed.tech/articles/keeping-a-human-in-the-loop-is-one-part-of-ai-agent-resilience-104729.md>)

Original publisher: [Read original article](<https://siliconangle.com/2026/10/28/human-in-the-loop-ai-agent-recovery-plans-aicybersecurityandresiliencesummit/>)

Author: Sloane Kali Faye

Published: 2026-10-28T17:00:23Z

Content type: news

Language: en

Sources: [SiliconANGLE](<https://devfeed.tech/sources/siliconangle.md>)

Topics: [Resilience](<https://devfeed.tech/topics/resilience.md>), [human in the loop agent](<https://devfeed.tech/topics/human-in-the-loop-agent.md>), [cyber resiliency](<https://devfeed.tech/topics/cyber-resiliency.md>), [observability](<https://devfeed.tech/topics/observability.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [trust](<https://devfeed.tech/topics/trust.md>)

Tags: [agentic-ai](<https://devfeed.tech/tags/agentic-ai.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agent](<https://devfeed.tech/tags/ai-agent.md>), [ai-agent-security](<https://devfeed.tech/tags/ai-agent-security.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-cybersecurity](<https://devfeed.tech/tags/ai-cybersecurity.md>), [ai-cybersecurity-and-resilience-summit](<https://devfeed.tech/tags/ai-cybersecurity-and-resilience-summit.md>), [ai-cybersecurity-and-resilience-summit-2026](<https://devfeed.tech/tags/ai-cybersecurity-and-resilience-summit-2026.md>), [ai-trust-and-resilience](<https://devfeed.tech/tags/ai-trust-and-resilience.md>), [aicybersecurityandresiliencesummit](<https://devfeed.tech/tags/aicybersecurityandresiliencesummit.md>), [aicybersecurityandresiliencesummit26eventpage](<https://devfeed.tech/tags/aicybersecurityandresiliencesummit26eventpage.md>), [autonomous-ai-agents](<https://devfeed.tech/tags/autonomous-ai-agents.md>), [bounded-autonomy](<https://devfeed.tech/tags/bounded-autonomy.md>), [cube-event-coverage](<https://devfeed.tech/tags/cube-event-coverage.md>), [cyber-recovery](<https://devfeed.tech/tags/cyber-recovery.md>), [cyber-resilience](<https://devfeed.tech/tags/cyber-resilience.md>), [cybersecurity](<https://devfeed.tech/tags/cybersecurity.md>), [dave-vellante](<https://devfeed.tech/tags/dave-vellante.md>), [human-in-the-loop](<https://devfeed.tech/tags/human-in-the-loop.md>), [krista-case](<https://devfeed.tech/tags/krista-case.md>), [news](<https://devfeed.tech/tags/news.md>), [resilience](<https://devfeed.tech/tags/resilience.md>), [secure-ai-foundations](<https://devfeed.tech/tags/secure-ai-foundations.md>), [security](<https://devfeed.tech/tags/security.md>), [security-infrastructure](<https://devfeed.tech/tags/security-infrastructure.md>), [thecube](<https://devfeed.tech/tags/thecube.md>), [thecube-research](<https://devfeed.tech/tags/thecube-research.md>)

## AI overview

Among 19 survey respondents with AI agents in production, none allowed high-risk actions without human approval, but only four of 25 had tested an AI or agent incident scenario. The interview highlights a recovery gap: restoring an agent does not undo records, transactions or workflows it changed, and 11 respondents reported gaps in reconstructing agent actions across systems. The agents' speed and scale make logging, monitoring, anomaly detection and forensics important if human controls fail; 12 of 25 respondents had not addressed anomalous-behavior detection. Organizations allowing independent low-risk actions were more likely to invest in monitoring and recovery and to test incidents, which the speakers present as a sign of more mature, bounded autonomy.

## Source excerpt

Keeping a human in the loop is one way enterprises can govern AI agents' actions. As those agents take on more work, security teams also need to consider how they would detect problems and recover from their effects. The relationship between AI trust and cyber resilience is becoming more complex as agents gain the ability [...] The post Keeping a human in the loop is one part of AI agent resilience appeared first on SiliconANGLE.