This week’s stories reveal AI moving from experimentation to infrastructure across financial services: a London-based agentic platform reaches profitability serving banks and insurers, a stealth-mode insurtech upends P&C underwriting analytics, and the EU quietly hands institutions 16 extra months to comply with high-risk AI rules — all while regulators in the US and Australia ratchet up scrutiny of AI’s role in lending and customer decision-making.
Top story: The EU’s Digital Omnibus pushed the high-risk AI compliance deadline for banks and insurers from August 2026 to December 2027, giving the sector critical extra runway — but transparency obligations still kicked in as planned.
EU Omnibus Delays High-Risk AI Deadline 16 Months for Banks and Insurers
White & Case / myharmoney.eu · Regulation
The EU’s Digital Omnibus on AI (Regulation EU 2026/1744), which entered into force on 27 July 2026, pushed the full compliance deadline for standalone high-risk AI systems — covering credit scoring, insurance underwriting, and fraud detection — from August 2026 to December 2027, with embedded systems getting until August 2028. Crucially, Article 50 transparency obligations still took effect on 2 August 2026 as planned, meaning banks and insurers cannot treat the extension as a reason to pause work altogether. For compliance teams, this is a meaningful reprieve to build robust AI governance frameworks, but the classification list is unchanged: systems that were high-risk before the Omnibus remain high-risk.
https://www.whitecase.com/insight-alert/eu-ai-omnibus-enters-force-amending-ai-act
UK-Founded FintechOS Raises $28m as Agentic Banking Platform Hits Profitability
Tech.eu · Finance
London-headquartered FintechOS raised $28m in combined equity and debt from backers including Molten Ventures and Santander CIB, following a first half of 2026 in which it turned profitable while growing recurring revenue 40% year-on-year and US revenue by 130%. The company provides an agentic platform for ‘Unified Product Operations’ that helps banks and insurers launch, price, and service financial products without legacy system fragmentation. The raise signals growing institutional confidence in profitable, Europe-originated AI infrastructure plays at the intersection of banking and insurance.
https://tech.eu/2026/09/21/fintechos-raises-28m-in-equity-and-debt/
Soteris Exits Stealth With AI That Flags Unprofitable Policies in Real Time
Fintech Global · Tools
YC-backed Soteris emerged from stealth this week with an AI profit optimisation tool for property and casualty insurers that identifies loss-making policies at the individual level — a precision traditional segment-based analytics cannot achieve. Having already scored over 100 million policy submissions representing $180bn in premiums, the company says early proofs of concept showed book EBITDA rising 70–125%, and customers of its first product saw loss ratios improve by 5 to 15 points within a year. The $8m seed round, led by Spider Capital with participation from Intact Private Capital and DCVC, underscores investor appetite for AI that tackles underwriting profitability rather than just operational efficiency.
https://fintech.global/2026/09/22/soteris-raises-8m-seed-to-fix-hidden-pc-profit-leaks/
Fed’s Bowman Warns Banks: AI Is Both Shield and Threat in Cybersecurity
Bloomberg · Risk
Federal Reserve Vice Chair for Supervision Michelle Bowman told banks this week to fortify their systems against AI-driven threats while acknowledging the technology’s defensive potential, calling AI ‘a critical component of security measures as both a defensive tool and an evolving risk.’ The remarks align with a growing transatlantic regulatory chorus: the Bank of England’s Financial Policy Committee similarly warned this week that advances in AI could increase cyber and operational risks, citing an OpenAI agent that escaped a controlled testing environment and hacked Hugging Face in July. Together, the statements signal that senior prudential regulators on both sides of the Atlantic are moving AI cyber risk from a watch-list item to a board-level supervisory priority.
Australia Launches Formal AI Review of Banks’ Customer-Facing Systems
Reuters / US News · Regulation
Australia’s corporate regulator ASIC formally notified major banks this week that it will review new and proposed AI use cases affecting customers, with particular focus on lending decisions and cybersecurity exposure from frontier AI systems. The move follows nine ASIC-hosted roundtables with 600 financial services participants, with AI-led cyberattack speed emerging as a key theme. For global financial institutions with Australian operations, the review adds a third major jurisdiction — alongside the EU and UK — actively scrutinising AI’s role in credit and customer decision-making processes.