AI GOVERNANCE CHAIN™← Back to the front page
Micro / Private · 3 to 10 people

The model that agreed its way toward a public failure

Presented by Mathieu K. Gouanou.

The complication

Days before launch, the founder noticed something no test suite had flagged: the assistant never disagreed. When trial users asserted things that were wrong, it validated them. When two instructions contradicted each other, it agreed with both. Nothing was broken in the code. The failure was in the behavior: an AI trained to please was about to face paying customers who needed it to be right, not agreeable.

The situation

A ten-person private company with no governance function, no compliance hire, and a customer-facing AI assistant at the center of its product. Like most teams of this size, it had assumed governance was a problem for later, for larger organizations with larger budgets.

The resolution

The team ran the free AI Sycophancy Test Battery over one afternoon. The model scored in the sycophantic band, failing hardest on flattery bias and capitulation under pushback. Two targeted corrections to the system instructions and a re-test moved it into the resistant band. The results were filed in the AI Sycophancy Risk Register™, and the product launched a week later with the evidence on record.

The outcome: a customer-facing failure prevented before launch, at near-zero cost, by a team with no governance budget.
The instruments behind this case: the free AI Sycophancy Test Battery and the AI Sycophancy Risk Register™ are in the Free Library. The full audit discipline is the Sycophancy Audit Toolkit in the paid catalog.

AI GOVERNANCE CHAIN™  |  © 2026 Mathieu K. Gouanou. All rights reserved.  |  Member, Harvard Business Review Advisory Council  |  Document standard: human-first AI governance. Anonymized case pattern; governance guidance, not legal advice.