---
id: "concept-basic-ai-failures"
type: "concept"
source_timestamps: ["§ Create intelligent failure protocols"]
source_url: "https://hbr.org/2026/02/how-to-foster-psychological-safety-when-ai-erodes-trust-on-your-team"
source_title: "How to Foster Psychological Safety When AI Erodes Trust on Your Team"
tags: ["failure-protocols", "process-improvement", "preventable-errors"]
related: ["concept-intelligent-ai-failures"]
definition: "Preventable errors, such as failing to verify AI outputs in domains where its limitations are already known, which should be avoided through better processes."
sources: ["adoption"]
sourceVaultSlug: "hbr-seg-adoption"
originDay: 9
articleStem: "hbr-cl-79-psychological-safety-ai-trust"
sourceUrl: "https://hbr.org/2026/02/how-to-foster-psychological-safety-when-ai-erodes-trust-on-your-team"
sourceTitle: "How to Foster Psychological Safety When AI Erodes Trust on Your Team"
---
# Basic AI Failures

In contrast to [[concept-intelligent-ai-failures]], **basic AI failures** are **preventable errors that occur in known contexts**. The canonical example is failing to verify AI outputs in a domain where the AI's limitations are *already well-documented and understood*.

Basic failures **do not generate new knowledge** — the team already knew (or should have known) the risk. Therefore they should not be celebrated; they should be **prevented** through better processes, checklists, and override protocols (see [[action-create-override-protocols]]). Treating a basic failure as if it were an intelligent one wastes the learning budget and erodes trust in the failure-protocol system.

The distinction between intelligent and basic failures is the discriminating logic inside the third pillar of the [[framework-ai-integration-principles|Psychological Safety Principles for AI Integration]]: *celebrate the boundary-pushing failures, engineer out the careless ones.*