During a financial crisis, even profitable firms face existential threats. The risk isn't from direct exposure to bad assets, but from a systemic "daisy chain" of distrust where counterparties refuse to pay their obligations, leading to a complete liquidity freeze that can bankrupt anyone.

Major companies like Amazon and financial service providers have integrated the SPLC's 'extremist' list into their compliance pipelines. In some cases, this authority is delegated, meaning a listing by the SPLC can automatically kill a transaction or account application as cleanly as an official government sanction.

The SPLC's list was adopted by financial firms partly due to a coordinated pressure campaign within its core community: nonprofits and their funders. The message was clear: screen donations using the SPLC list or face social and financial consequences, effectively bootstrapping its data product into the financial supply chain.

Financial institutions are required to file Suspicious Activity Reports (SARs) with the government. These detailed memos, funded by the banks, often serve as pre-written indictments for prosecutors, who can sometimes directly copy the narrative into a formal legal complaint, effectively outsourcing investigative work.

Instead of building bespoke systems, banks buy 'data products' from screening vendors to check against lists like the government's OFAC list. These vendors bundle official sanctions lists with private ones, such as the SPLC's 'Extremist files,' effectively creating a market for outsourced compliance decision-making.

While many focus on AI for consumer apps or underwriting, its most significant immediate application has been by fraudsters. AI is driving an 18-20% annual growth in financial fraud by automating scams at an unprecedented scale, making it the most urgent AI-related challenge for the industry.

Large-scale fraud operates like a business with a supply chain of specialized services like incorporation agents, mail services, and accountants. While some tools are generic (Excel), graphing the use of shared, specialized infrastructure can quickly unravel entire fraud networks.

A defender's key advantage is their massive dataset of legitimate activity. Machine learning excels by modeling the messy, typo-ridden chaos of real business data. Fraudsters, however sophisticated, cannot perfectly replicate this organic "noise," causing their cleaner, fabricated patterns to stand out as anomalies.

Purely model-based or rule-based systems have flaws. Stripe combines them for better results. For instance, a transaction with a CVC code mismatch (a rule) is only blocked if its model-generated risk score is also elevated, preventing rejection of good customers who make simple mistakes.