---
id: "question-measuring-sabotage"
type: "open-question"
source_timestamps: ["§ Engage Your People"]
tags: ["metrics", "hr", "risk"]
related: ["concept-ai-sabotage", "claim-human-bottleneck"]
resolutionPath: "Development of telemetry or HR feedback loops that differentiate genuine technical bugs from intentional employee metrics tampering."
source_url: "https://hbr.org/2026/01/match-your-ai-strategy-to-your-organizations-reality"
source_title: "Match Your AI Strategy to Your Organization's Reality"
sources: ["spine"]
sourceVaultSlug: "hbr-seg-spine"
originDay: 1
articleStem: "hbr-sig-55-match-ai-strategy-to-reality"
sourceUrl: "https://hbr.org/2026/01/match-your-ai-strategy-to-your-organizations-reality"
sourceTitle: "Match Your AI Strategy to Your Organization’s Reality"
---
# How can organizations preemptively detect AI sabotage?

**Open question.** If **10%** of employees engage in [[concept-ai-sabotage]] (per [[claim-human-bottleneck]]), how can an organization detect it *before* it damages ROI?

**Resolution path.** Develop telemetry or HR feedback loops that differentiate genuine technical bugs from intentional metrics tampering or deliberately low-quality outputs. Open because the source asserts the prevalence but offers no detection method — and, per the enrichment overlay, the 10% figure itself is unverified, so any detection system would also need to establish a credible base rate.