---
id: "claim-policing-ai-impossible"
type: "claim"
source_title: "Don't Let AI Slop Muck Up Your Company's Processes"
source_url: "https://hbr.org/2026/06/dont-let-ai-slop-muck-up-your-companys-processes"
source_timestamps: ["§ How to Deal with Knowledge Decay?", "¶20"]
tags: ["governance", "shadow-it"]
related: ["framework-four-steps-knowledge-decay"]
speakers: ["Matthias Holweg", "Thomas H. Davenport"]
confidence: "high"
testable: true
sources: ["execution"]
sourceVaultSlug: "hbr-seg-execution"
originDay: 8
articleStem: "hbr-sig-54-ai-slop-processes"
sourceUrl: "https://hbr.org/2026/06/dont-let-ai-slop-muck-up-your-companys-processes"
sourceTitle: "Don’t Let AI Slop Muck Up Your Company’s Processes"
---
# Policing AI usage at work is virtually impossible

**Claim:** Organizations cannot effectively ban or police employees' use of generative AI. Surveys indicate that more than half of workers conceal their use of the technology. Leaders must therefore focus on process redesign and structural constraints — such as structured data inputs — rather than prohibition.

This premise motivates the entire [[framework-four-steps-knowledge-decay]] and specifically [[action-restrict-unstructured-inputs]]; the enforcement gap it creates is [[question-detecting-ai-content]].

**Confidence:** high (author) / *supported as a practical governance challenge; the strong formulation needs more evidence* (enrichment). The US national-lab study shows quiet risk-taking — workers make dozens of micro-decisions about AI use without guidance and with 'no clear governance policies.' But the specific 'more than half conceal their use' statistic isn't visible in the cited sources, and NIST/industry frameworks recommend acceptable-use policies, content labeling, and detection tools — implying partial control is feasible, so 'virtually impossible' is an overstatement. **Testable:** yes.


## Related across articles
- [[claim-governance-targets-wrong-problem]]
- [[concept-ai-knowledge-hiding]]