# ACPR fines Crédit Agricole du Languedoc €1.5m over gaps in automated monitoring found with its AI tool

Source: https://www.bankingnewsai.com/ai-incidents/credit-agricole-languedoc-2022-acpr-monitoring-scenarios
Last updated: Oct 21, 2026

Caisse régionale de Crédit Agricole Mutuel du Languedoc · Governance finding · severity 3 of 5

On 1 December 2022 the sanctions committee of France's banking supervisor, the ACPR, reprimanded the Caisse régionale de Crédit Agricole Mutuel du Languedoc and fined it €1.5 million for anti-money-laundering failings. Per the ACPR, the regional bank's automated transaction-monitoring system relied on an incomplete set of scenarios that could not detect certain types of atypical transactions: its cash-deposit scenario flagged only weekly deposits above €14,000 when average deposits were €400–700, and it had no scenario for small payments that could signal terrorist financing or for mismatches between income and inflows. The inspection was the first ACPR on-site inspection to use the supervisor's own AI tool, LUCIA, which processed data on 750 million payment transactions. The bank's monitoring was scenario-based and not described as AI.

## Key facts

|  |  |
| --- | --- |
| Date | Dec 1, 2022 (disclosed Dec 7, 2022) |
| Institution | Caisse régionale de Crédit Agricole Mutuel du Languedoc |
| Country | France |
| AI system | Scenario-based automated anti-money-laundering transaction monitoring (scenarios such as S03 for cash deposits); not described as AI. The inspectors separately used the supervisor's own AI tool, LUCIA. |
| AI? | Automated system not described as AI — The system was automated, but neither the institution nor a regulator described it as AI; it is listed because it is routinely discussed as an AI failure. |
| Category | Regulatory finding (Governance finding) |
| Status | Regulator finding |
| Severity | 3 of 5 — A regulatory finding with a reprimand and a €1.5 million fine: the regulator found a significant failure in the bank's automated transaction monitoring that left certain atypical transactions undetected; no customer loss was reported. |

## What happened?

The ACPR inspected the Caisse régionale de Crédit Agricole Mutuel du Languedoc (CRCAM Languedoc) from 29 June 2020 to 15 January 2021. The regional bank, part of the Crédit Agricole group, had about 2,700 staff and about one million customers in the Aude, Gard, Hérault and Lozère départements, where its market share exceeded 30%. On 15 October 2021 the ACPR's supervisory college opened disciplinary proceedings on the basis of the inspection report. [1]

Per the decision, the inspection was the first ACPR on-site inspection to use LUCIA ("Logiciel à l'Usage du Contrôle assisté par l'Intelligence Artificielle"), the supervisor's AI tool, after the bank transferred 540 GB of data covering 750 million payment transactions made by customers between January 2018 and June 2020. The bank argued it had not been told in time how the data would be used and that the tool's use was disproportionate and harmed its rights of defence. The committee rejected the argument, holding that inspectors "ne sont pas tenus de faire connaître aux organismes contrôlés les caractéristiques des logiciels qu'ils peuvent être, le cas échéant, conduits à utiliser" ("are not required to disclose to the inspected institutions the characteristics of the software they may use"; our translation). [1][3]

On the monitoring complaint (grief 2), the committee found that a group scenario applied without adaptation, S03, detected only weekly cash deposits by individuals above €14,000 and above five times the previous three-month average, whereas average deposits were between €400 and €700; per the inspectors it excluded 90% of non-professional account holders. No scenario was adapted to small payments that could relate to terrorist financing, and none tracked consistency between customers' known income and the credits to their accounts. The committee said human vigilance could only complement the alerts of an automated system in a bank of this size. A separate complaint about a cheque scenario, S17, was dismissed because the bank had tested it for two months in 2018 and then dropped it. [1]

The committee concluded: "le dispositif de suivi et d'analyse des opérations de la clientèle de la CRCAM Languedoc reposait sur un ensemble de scénarios incomplet, ce qui ne la mettait pas en mesure de détecter certains types d'opérations atypiques" ("the system for monitoring and analysing customers' transactions rested on an incomplete set of scenarios, which did not enable it to detect certain types of atypical transactions"; our translation). It also upheld know-your-customer gaps, 17 failures to carry out enhanced reviews and 9 failures to file suspicious-activity reports with Tracfin, but treated these as isolated failings. It said the improvements expected from the bank's new AI-based detection tool did not change the finding. [1][2]

## Timeline

- Jun 29, 2020 — The ACPR begins an on-site inspection of CRCAM Languedoc, using its AI tool LUCIA for the first time on such an inspection; the inspection runs to 15 January 2021. [1]
- Oct 15, 2021 — The ACPR supervisory college decides to open disciplinary proceedings against the bank. [1]
- Nov 14, 2022 — Public hearing before the sanctions committee; the college's representative seeks a reprimand and a €2 million fine. [1]
- Dec 1, 2022 — The sanctions committee issues a reprimand and a €1.5 million fine and orders publication under the bank's name for five years. [1]
- Dec 7, 2022 — The ACPR publishes the decision and a press release. [2]

## What was the impact?

- Financial: €1.5 million fine and a reprimand
- Customers: No customer loss reported; the decision does not quantify the transactions the scenarios missed.
- Other: The regulator found a significant failure of a component of the bank's anti-money-laundering system, illustrated by individual files that should have led to enhanced reviews or reports to Tracfin. [1][2]

## How did Caisse régionale de Crédit Agricole Mutuel du Languedoc respond?

Before the committee the bank argued that the way LUCIA was used made the procedure irregular, stressed the "caractère inédit et disproportionné" (unprecedented and disproportionate nature) of the means LUCIA used, said the shortcomings found were marginal in light of those means, cited its anti-money-laundering remediation (more financial-security staff and an early role in a group financial-crime detection programme using innovative technologies) and asked for the decision to be published anonymously or not at all. We found no separate public statement by the bank as of 2026-10-21 (searched the ACPR site, French and trade press, and the bank's website). [1]

## What was the outcome?

The committee imposed a reprimand and a €1.5 million fine, less than the €2 million sought, and ordered the decision published in the ACPR register under the bank's name for five years. The decision could be appealed within two months; as of 2026-10-21 we found no published appeal ruling. [1][2]

## FAQ

### Why was Crédit Agricole du Languedoc fined by the ACPR?

The ACPR's sanctions committee found in December 2022 that the regional bank's automated transaction monitoring rested on an incomplete set of scenarios that could not detect certain atypical transactions, alongside know-your-customer gaps, missed enhanced reviews and nine missing suspicious-activity reports. It imposed a reprimand and a €1.5 million fine.

### Was the bank's monitoring system AI?

No. The decision describes a scenario-based automated monitoring system with fixed thresholds, such as a €14,000 weekly cash-deposit trigger. The AI in this case belonged to the supervisor: the ACPR's inspectors used their LUCIA tool to analyse data on 750 million payment transactions.

### Did the bank challenge the regulator's use of AI?

Yes. The bank argued that it was not told in time that LUCIA would analyse almost all of its data and that the tool's use was disproportionate and unfair. The sanctions committee rejected this, saying inspectors need not disclose the characteristics of their software and that the bank's defence rights were not harmed.

### Was any money lost because of the monitoring gaps?

No loss figure was published. The decision illustrates the gaps with individual customer files whose transactions should have triggered an enhanced review or a suspicious-activity report to Tracfin, France's financial intelligence unit, but it does not estimate the value of transactions the scenarios missed.

## Sources

1. [Décision de la Commission des sanctions, procédure n° 2021-05, Caisse régionale de Crédit Agricole Mutuel du Languedoc](https://acpr.banque-france.fr/system/files/import/acpr/media/2022/12/07/20221207_decision_crcam_languedoc.pdf) — ACPR Commission des sanctions, Dec 7, 2022 (official): "Il résulte de tout ce qui précède qu’au moment du contrôle sur place, le dispositif de suivi et d’analyse des opérations de la clientèle de la CRCAM Languedoc reposait sur un ensemble de scénarios incomplet, ce qui ne la mettait pas en mesure de détecter certains types d’opérations atypiques (grief 2)."
2. [La Commission des sanctions de l’ACPR sanctionne la Caisse régionale de Crédit agricole mutuel du Languedoc](https://acpr.banque-france.fr/fr/communiques-de-presse/la-commission-des-sanctions-de-lacpr-sanctionne-la-caisse-regionale-de-credit-agricole-mutuel-du) — Autorité de contrôle prudentiel et de résolution (ACPR), Dec 7, 2022 (official): "Par une décision du 1er décembre 2022, la Commission des sanctions de l’ACPR a prononcé un blâme et une sanction pécuniaire de 1,5 million d’euros à l’encontre de la Caisse régionale de Crédit agricole mutuel du Languedoc (CRCAM Languedoc)."
3. [France Fines Credit Agricole After Analyzing 750 Million Transactions Using AI](https://www.acams.org/en/news/france-fines-credit-agricole-after-analyzing-750-million-transactions-using-ai) — ACAMS moneylaundering.com, Dec 8, 2022 (news): "France's financial regulator fined Credit Agricole €1.5 million for anti-money laundering violations in its first enforcement action stemming from the use of an artificial intelligence tool that allowed examiners to analyze hundreds of millions of the bank's transactions and other data."

## Related incidents

- [ACPR fines Axa Banque €1m over unsuitable automated transaction-monitoring scenarios](https://www.bankingnewsai.com/ai-incidents/axa-banque-2023-acpr-monitoring-scenarios)
- [FCA fines Metro Bank £16.7m after its automated transaction monitoring missed 60m transactions](https://www.bankingnewsai.com/ai-incidents/metro-bank-2024-fca-automated-transaction-monitoring)

All AI incidents at banks: https://www.bankingnewsai.com/ai-incidents

---

Canonical page: https://www.bankingnewsai.com/ai-incidents/credit-agricole-languedoc-2022-acpr-monitoring-scenarios
Part of [BankingNewsAI](https://www.bankingnewsai.com/) — the register of AI incidents at banks: https://www.bankingnewsai.com/ai-incidents. Markdown versions of every reference page: append `.md` to the page URL; index at https://www.bankingnewsai.com/llms.txt.