Promise... but without guardrails 3 AI was nowhere in the 2024 Compass . In 2026 it is everywhere. It is the #2 overall trend (up from dead last in 2024) and AI-related risk is the #1 challenge. Respondents of course see its practical value — notably in underwriting, back-office efficiency, client interaction and detecting fraud/AML-KYC. There is a strong consensus that its strongest use-case, for now at least, is in credit appraisal or approval. However, the same technology also raises very significant concerns around fraud, bias, transparency, data privacy or misuse, and the increased exclusion of digitally invisible clients . There is a profound sense (a message clearly conveyed at the IF25 conference and in the proposals for IF26) that the proliferation and growing severity of threats by AI is far outpacing regulators’ or other ecosystem players’ ability to understand and respond . So the real test here of whether the promise of AI can be grasped and its role contained is a question of institutional readiness and governance. 3. Five themes I think there will ultimately be a positive impact from AI, but there will be a long ‘winter’ in which AI causes far more problems than it solves for this sector. Timothy Ogden, Managing Director, Financial Access Initiative