In April, the prestigious law firm Sullivan & Cromwell issued an apology after submitting a bankruptcy court filing containing multiple errors generated by artificial intelligence (AI). While the incident drew attention, it also underscored the growing reliance on AI tools within top-tier legal practice. The episode has sparked renewed debate about the role of AI in law, especially as legal professionals and regulators weigh the technology’s potential benefits and risks.
Under longstanding Supreme Court precedent, the right to legal counsel applies only in criminal cases, leaving civil litigants without a guaranteed attorney. State laws broadly prohibit nonlawyers, including AI-powered chatbots, from representing individuals in court. As a result, approximately 75 percent of the more than 15 million annual civil cases involve at least one party without legal representation. Legal experts see this gap as a significant opportunity for AI applications.
David Freeman Engstrom, a legal scholar at Stanford Law School, notes that much of legal practice involves applying established rules to facts, a task well-suited to large language models (LLMs). He points out that many civil cases, such as debt collections or evictions, tend to be routine and noncontroversial. His research on California debt-collection cases found nearly half contained uncaught errors, suggesting that better representation—potentially supported by AI—might change outcomes for some defendants.
However, current unauthorized practice of law statutes impose barriers to deploying AI assistance, even for routine matters. These regulations do not distinguish among case types and typically prevent AI services from providing substantive legal help. Advocates argue that revising these restrictions could enhance access to justice by reducing costs and empowering litigants.
Beyond improving affordability, AI has the potential to increase efficiency in the justice system. Traditional adversarial proceedings depend on the skill and resources of lawyers to present facts, but inequalities in advocacy can affect case outcomes. AI tools can drastically reduce the marginal cost of legal work and challenge the notion that legal truth hinges on quality of representation. Some experts suggest that well-designed AI systems could, in principle, adjudicate certain issues independently, highlighting that some nonlegal aspects of lawyering—such as investigation or cultivating relationships with judges—do not directly bear on legal merits.
The legal profession has expressed concerns about competition from AI, paralleling similar tensions seen in medicine, where AI has outperformed doctors in some diagnostic tasks. Licensing requirements and professional norms act as protective measures. Judicial pushback against AI errors, such as a recent reprimand of a prosecutor in North Carolina for failing to correct AI-generated false citations, reflects ongoing wariness. Nonetheless, observers emphasize that the key takeaway should be the importance of rigorous review, regardless of whether work was done by humans or machines.
Wider AI adoption may lead to increased litigation, as lower transaction costs encourage enforcement of rights once deemed too costly to pursue. Some worry this could overwhelm courts with trivial claims. Others respond that this phenomenon exposes the overregulation of society and that the appropriate remedy lies in legislative reform, not restricting access to legal tools. Courts could also employ AI themselves to filter baseless filings more efficiently.
As AI legal assistants become more capable and cost-effective, they may democratize access to legal services, transforming what was once a privilege of the wealthy into a broad public resource. The legal profession’s deepest unease appears not to be about AI’s limitations but rather the possibility that AI will soon become sufficiently effective to reshape the practice of law itself.