---
id: "claim-unempathetic-rollouts-sabotage"
type: "claim"
source_timestamps: ["§ When Empathy Goes"]
tags: ["insider-threat", "sabotage", "zero-sum-dynamics"]
related: ["concept-workslop", "entity-writer", "contrarian-ai-sabotage"]
speakers: ["Jamil Zaki"]
confidence: "high"
testable: true
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"
---
# Unempathetic AI rollouts provoke active employee sabotage

**Claim:** Unempathetic AI rollouts provoke active employee sabotage. **(Confidence: high per source; see caveats)**

When AI is forced on a workforce without empathy or trust, employees perceive a [[prereq-zero-sum-environment]] and take natural — albeit unethical — steps to protect themselves. An enterprise-AI adoption survey by [[entity-writer]] found that nearly **a third** of all employees, and a staggering **44% of Gen Z** workers, admit to actively sabotaging their company's AI strategies.

Methods of sabotage include feeding sensitive information to unauthorized models and intentionally tampering with outputs to make the AI appear less effective. This escalates beyond passive [[concept-workslop-d42]] and indicates that ignoring psychological safety creates severe operational *and security* risks. This claim anchors the contrarian reframe [[contrarian-ai-sabotage]] and the unresolved [[question-sabotage-prevention]].

**Enrichment / confidence:** Conceptually well supported — organizational research on counterproductive work behavior shows perceived injustice, threat, and disrespect can yield deliberate sabotage; insider-threat literature notes non-compliant use and data exfiltration in low-trust rollouts. BUT the specific Writer statistics (≈one-third; 44% of Gen Z) are vendor-produced, self-reported, and not independently verifiable; self-reports may conflate exploratory/non-compliant behavior with genuine malicious sabotage. Treat prevalence as likely overstated or context-specific.


## Related across articles
- [[claim-active-sabotage]]
- [[contrarian-ai-sabotage]]