---
id: "concept-standard-rai-approach"
type: "concept"
source_timestamps: ["§ Three Flaws of the Standard Approach", "¶1"]
source_url: "https://hbr.org/2026/05/what-are-your-companys-ai-nightmares"
source_title: "What Are Your Company's AI Nightmares?"
tags: ["ai-governance", "compliance", "corporate-policy"]
related: ["claim-standard-rai-too-slow", "framework-standard-rai-model", "concept-agentic-ai-governance-gap"]
definition: "A top-down AI governance model that translates abstract ethical values into enterprise-wide policies enforced by centralized risk boards."
sources: ["governance"]
sourceVaultSlug: "hbr-seg-governance"
originDay: 7
articleStem: "hbr-cl-82-ai-nightmares"
sourceUrl: "https://hbr.org/2026/05/what-are-your-companys-ai-nightmares"
sourceTitle: "What Are Your Company’s AI Nightmares?"
---
# Standard Responsible AI (RAI) Approach

The conventional methodology enterprises adopted to manage AI ethical risks before and during the early generative-AI boom. It relies on a **top-down, values-first architecture**, formalized in [[framework-standard-rai-model]]:

1. The organization articulates abstract AI ethics **values** (e.g., fairness, privacy, transparency, accountability, safety).
2. Those values are **translated into enterprise-wide procedures** (e.g., bias checking, filtering sensitive data).
3. The procedures are **enshrined in a formal, enterprise-wide policy**.
4. The policy is **implemented** across the organization.
5. Enforcement is delegated to a **centralized Responsible AI board** (or an existing risk board) that handles escalations of high-risk AI cases.

Blackman ([[entity-reid-blackman]]) argues this approach is **fundamentally broken** — see the quote [[quote-standard-approach-broken]]. It is bottlenecked by the C-suite, takes **upwards of a year** to implement (see [[claim-standard-rai-too-slow]]), relies on inscrutable, jargon-heavy policy language (the [[quote-tower-of-babel]] problem), and fails to define concrete success metrics (the root of [[claim-values-wrong-start]]). Its failure becomes acute in the era of agentic AI — see [[concept-agentic-ai-governance-gap]].

**Enrichment note:** Blackman's own Substack essay and LinkedIn writing directly support the "fundamentally broken" verdict. Broader industry commentary echoes the concerns about *speed* and *bureaucracy* but is often more measured — many practitioners would not go so far as to call the model "broken," and some argue policy-first approaches can be made agile (see [[contrarian-corporate-optimism-liability]] and the counter-perspectives in [[_AGENT_PRIMER]]). Understanding this model presupposes [[prereq-corporate-governance-structures]].


## Related across articles
- [[framework-ai-risk-oversight]]
- [[framework-board-evolution-pyramid]]