Curated Insight: EU Parliament report on AI in FS

The European Parliament adopted a new report on the impact of artificial intelligence on the financial sector recently. You can read the full text here.

The report covers what is now familiar to us: AI must be reliable, explainable, and subject to meaningful human oversight. It highlights bias, hallucinations, low quality data, and the dangers of black box models.

It reiterates that under the EU AI Act, credit assessments and insurance pricing are high-risk use cases.

However, the report also frames AI as a clear opportunity for the sector. It specifically calls out potential benefits, including better fraud detection.

The recommendation to supervisors is also interesting. It emphasises the need for continuous monitoring of governance and risk management frameworks. This signals a move away from simple tick-box compliance.

Why this matters

The report reinforces that reliability is not optional. If we use models that fabricate sources or hallucinate facts, given the consistent warnings that supervisors have provided on this, it is a failure that we should have foreseen.

The shift towards testing and documentation suggests that supervisors will look for evidence that we understand our models, not just a policy document. While still important of course, compliance is not the end goal.

Disclaimer: The info in this article is not legal advice. It may not be relevant to your circumstances. It was written for specific contexts within banks and insurers, may not apply to other contexts, and may not be relevant to other types of organisations.