Major Wall Street banks are pushing large law firms to slash their fees, arguing that artificial intelligence tools can absorb significant portions of traditional legal work.
The Financial Times reported that investment banks are increasingly questioning the value of expensive legal bills when AI systems such as ChatGPT can generate answers on demand.
This pressure arrives despite a high-profile cautionary tale involving a gaming chief executive who asked ChatGPT to provide legal grounds for avoiding a $250 million bonus payment.
That executive lost the case, yet financial institutions appear undeterred in their confidence that AI can safely handle consequential legal matters.
The push sits awkwardly alongside signals from ILTACON, an industry conference held last week, where clients were reportedly telling law firms they wanted outcomes at virtually any price.
Critics argue that the monied interests backing AI have burned close to one and a half trillion dollars in expenses to generate around $200 billion, a ratio that strains conventional investment logic.
The finance sector’s faith in AI appears tied less to performance evidence and more to a necessary article of belief for institutions that have committed heavily to the technology on their balance sheets.
Without that faith, the uncomfortable reality of hallucinations, boilerplate drawn from unrelated deals, and suboptimal legal research becomes far harder to ignore.
Legal technology experts are increasingly acknowledging that so-called human-in-the-loop frameworks are insufficient when they assume a senior partner will review 200 pages of agentic logs at the backend of a process.
The legal industry, observers argue, needs AI to function as a cyborg rather than a robot, working in genuine conjunction with human judgment rather than replacing it at every stage.
AI is also having a counterintuitive effect on lawyer productivity, with early indications suggesting it is increasing billable hours rather than reducing them, as AI surfaces research paths that humans then pursue to improve their work product.
The compression of document review time is itself a problem, since the hours previously spent working through complex agreements allowed lawyers to experience the strategic rethinks and epiphanies that improve final output.
AI delivers what one observer described as the median acceptable answer by design, smoothing prior results into a form that may provide a useful starting point but cannot serve as a finished work product.
Flat fees for repeatable tasks have been proposed as one workable compromise, offering clients cost predictability without forcing law firms to surrender revenue entirely to AI-driven discounting.
Banks pressing for deep fee cuts and AI substitution may find themselves confronting the limits of that strategy once litigation arising from today’s transactions begins in earnest.