---
id: "question-sabotage-prevention"
type: "open-question"
source_timestamps: ["§ When Empathy Goes"]
tags: ["security", "trust-dynamics", "insider-threat"]
related: ["claim-unempathetic-rollouts-sabotage"]
resolutionPath: "Case studies on organizations that successfully remediated high levels of AI sabotage through cultural interventions rather than punitive surveillance."
sources: ["adoption"]
sourceVaultSlug: "hbr-seg-adoption"
originDay: 9
articleStem: "hbr-edu-42-empathetic-leadership-ai-adoption"
sourceUrl: "https://hbr.org/2026/04/empathetic-leadership-can-make-or-break-ai-adoption"
sourceTitle: "Empathetic Leadership Can Make or Break AI Adoption"
---
# How can organizations detect and prevent AI sabotage without further eroding trust?

**Open question:** How can organizations detect and prevent AI sabotage without further eroding trust?

The source reveals that nearly a third of employees (and 44% of Gen Z) admit to sabotaging corporate AI strategies — feeding sensitive info to unauthorized models or tampering with outputs (see [[claim-unempathetic-rollouts-sabotage]]). The bind: implementing strict surveillance or punitive measures to stop sabotage would likely *further* destroy the [[prereq-psychological-safety-d42]] and empathy required for successful adoption. The text does not detail operational mechanisms for securing AI models against insider sabotage while simultaneously building trust.

**Resolution path:** Case studies on organizations that remediated high AI-sabotage levels through *cultural* interventions rather than punitive surveillance.

**Enrichment note:** Because the underlying prevalence stats are unverified vendor data, part of resolving this is establishing the *true* base rate of malicious (vs. merely non-compliant) behavior before designing interventions.