In late 2025, a top-five Moroccan bank — anonymised here as 'Bank M' — received an AML-related sanction from Bank Al-Maghrib. The fine itself, roughly 45 million dirhams, was immaterial to the income statement. The reputation damage was not. Over the following 30 days, Bank M's Harch reputation score fell 18 points, the steepest single-incident decline we recorded in the Moroccan banking sector in 2025. This is the anatomy of that decline, and the lessons it leaves.
The 30-day score decline
Our score combines media sentiment (weighted 40%), AI-engine citation sentiment (25%), social-platform conversation tone (20%), and expert-panel review (15%). The chart below tracks Bank M's score across the 30 days following the public disclosure of the sanction.
Two patterns matter. First, 55% of the total decline happened in the first 72 hours — before the bank's formal response was published. Second, the score flatlined at 60 from day 21 onwards, indicating a new, lower equilibrium rather than a recovery. Six months later, the score had recovered only 4 points.
The cost breakdown
We estimate the total reputation cost at roughly 380 million dirhams over 12 months — over eight times the headline fine. The breakdown below is modelled from disclosed retail-deposit flows, internal customer-acquisition cost benchmarks, and NPS survey data shared with us on background.
What the response got right
Bank M did three things well. The CEO issued a video statement within 48 hours acknowledging the finding without minimising it. The bank published a remediation plan with named owners and deadlines — the three-anchor structure we recommend for any crisis communication. And it briefed its top 200 corporate clients individually within the first week, pre-empting relationship-manager calls they could not otherwise have handled.
“The fine we could absorb. What we could not absorb was the silence — every hour without a clear, owned narrative was an hour the narrative was written for us.”— Chief Communications Officer, Bank M (anonymised)
What the response got wrong
Three failures stand out. The bank had no pre-prepared AML holding statement — the legal team drafted from scratch and the 48-hour delay cost an estimated 8 points of score. The bank did not monitor AI engines; on day 4, ChatGPT answers to 'is Bank M safe' were already citing the sanction with no counter-narrative. And the bank's social listening covered only French and Arabic MSA, missing the Darija conversation on TikTok and X that drove the sharpest sentiment drop on day 3.
The AI-engine tail
The most enduring cost was the AI-engine tail. Twelve weeks after the incident, ChatGPT and Perplexity were still citing the sanction in answer to evaluative prompts, in 73% and 81% of cases respectively. The bank's counter-narrative — the remediation plan, the clean subsequent audit — had not entered the training surface in sufficient density to displace the original incident. We estimate the tail extended the score recovery from a potential 9 months to 18 months.
Lessons for 2026
- Pre-draft holding statements for the five most likely crisis scenarios. The drafting time, not the messaging, is what costs points.
- Monitor AI engines from day zero of an incident. A counter-narrative seeded in the first 72 hours reduces the 12-week tail by up to 60%.
- Cover Darija. French and Arabic MSA monitoring misses the fastest-moving sentiment channel in Morocco.
- Brief corporate clients individually in week one. The relationship-manager channel is the most efficient stabiliser.
- Model the reputation cost — not just the fine — in the board's incident scenario planning.
The headline number from this case is not the fine. It is the 8.4× multiplier between the disclosed penalty and the modelled reputation cost. That multiplier is the figure boards should plan against.
Harch Atelier's Crisis Post-Mortem service reconstructs incident timelines, scores the 30-day reputation impact, and benchmarks the response against the 72-hour best-practice window.