AI Adoption Grows, But So Do Fraud Teams
A new report from SEON finds that while AI use is nearly universal among fraud and anti-money laundering (AML) professionals, both headcount and budgets for these teams are increasing. The survey of over 1,000 leaders suggests that fragmented systems and regulatory complexity require continued human oversight. This indicates that AI is currently augmenting, rather than replacing, skilled fraud and compliance personnel.
The SEON report's headline figures show 98% of fraud and AML teams now use AI, with 83% expecting budget increases and 94% planning to hire more staff in 2026. The core challenge isn't AI's efficacy—95% are confident it works—but operational fragmentation; only 47% of firms have fully integrated fraud and AML workflows, and 80% find it difficult to get a unified view of their data. This human-in-the-loop reality is driving a shift toward agentic AI architectures, moving beyond passive generative AI to autonomous, goal-oriented systems. In these designs, an orchestrator or "supervisor" agent assigns specialized tasks—like data retrieval or anomaly analysis—to other agents, mimicking an investigative team to handle dynamic, multi-step compliance workflows in areas like Know Your Customer (KYC). This approach requires a composable backend architecture, introducing an "agentic layer" via APIs on top of legacy systems. For insurtechs, this pattern is directly applied to claims and underwriting. One Nordic insurer automated 70% of claims document interpretation using an AI solution that combined optical character recognition (OCR) and natural language processing (NLP) to structure incoming data. Another Dutch motor insurer went further, using an AI agent to achieve 91% automation for eligible claims decisions, cutting processing time by 46% and boosting its Net Promoter Score by 9%. The technical leadership challenge for a Principal Engineer is to influence this transition without direct authority, guiding teams through complex architectural tradeoffs and setting the standards for integrating these systems. This involves championing API-first design to break down data silos and leveraging open-source LLM orchestration frameworks like LangChain or Haystack to build reproducible, stateful AI workflows that can be audited and governed. Venture capital is aggressively funding this shift, with global insurtech investment hitting $5.08 billion in 2025, a 19.5% year-over-year increase. AI-native companies commanded two-thirds of all 2025 funding. Recent major rounds in early 2026, like AI-native carrier Corgi's $108 million raise, signal that investors are backing full-stack, AI-driven platforms that can modernize core processes like underwriting and claims from the ground up. Criminals are also leveraging AI for more sophisticated attacks, with AI-assisted document forgery and deepfakes growing rapidly. One report noted that digitally-presented media was 300% more likely to be AI-generated or altered in 2025 compared to the prior year. This creates an arms race, where fraud teams must continuously evolve their AI defenses to counter AI-powered threats, further justifying the increased headcount and budgets.
