Field Note: Even Biglaw Gets The AI Hallucination Blues

Even elite firms can get burned by AI hallucinations. In this field note, Ron Drescher breaks down the recent Sullivan & Cromwell filing controversy, where an emergency brief reportedly contained multiple citation and quotation errors that opposing counsel exposed first. The lesson is not that one firm slipped—it’s that polished AI output can create false confidence in any lawyer, especially when they're under severe deadline pressure.

Ron connects the story to his earlier Confession of an AI Hallucinator episode (where he confessed to sending out a memo containing hallucinated cases) and explains why time-stressed emergency filings are fertile ground for hallucination mistakes. He then pivots to a practical alternative: using AI as a research guide, not a research substitute.

Why Schedule A Matters

The recent Sullivan & Cromwell letter regarding AI-related errors in a court filing has generated significant discussion in the legal world, but the real story may be found in Schedule A. The schedule catalogs a striking number of citation errors, quotation issues, and other inaccuracies that appeared throughout a filed motion. These were not isolated typos or one-off mistakes. They were varied, recurring, and spread across the document in ways that raise serious questions about how AI-generated legal drafting is being used and reviewed.

For lawyers, Schedule A is an important reminder that polished language can create a false sense of reliability. AI tools may produce persuasive prose and professional-looking citations, but appearance is not authority.

The deeper lesson is not that lawyers should avoid AI altogether. Used properly, AI can be extremely helpful for brainstorming arguments, organizing facts, generating search strategies, summarizing documents, and improving workflow. But Schedule A underscores the danger of using AI as a substitute for trusted legal research or as an unchecked drafting engine for court submissions.

My view is simple: AI works best as a research guide and productivity assistant, not as the final source of legal authority. Schedule A may become a landmark example of why that distinction matters.

Instead of relying on AI to generate authorities directly, Ron proposes using AI to create multiple Boolean search strategies, help navigate Westlaw/Lexis/Bloomberg features, and improve the research process while keeping lawyers anchored to real databases, real cases, and real citations.

The episode also introduces Ron’s “airport metaphor” for after-the-fact AI verification: if the promised shortcut requires hours of extra checking after the brief is drafted, maybe the traditional route would have been faster all along. If verification occurs only after a brief has been drafted, the cleanup process can become so extensive that the promised efficiency gains may disappear entirely. In some matters, it may truly have been easier to “drive than fly.”

Key Topics Covered

• Why AI hallucinations are not just a solo/small-firm problem

• What happened in the Sullivan & Cromwell filing controversy

• Why emergency motions and deadline pressure increase hallucination risk

• The danger of polished but false AI output

• Why Ron is skeptical of AI-as-research-substitute workflows

• Using AI to generate smarter Boolean searches

• Using AI to help master legal research tools you already pay for

• Building briefs from verified authority rather than unverifiable AI citations

• The airport metaphor for inefficient AI verification workflows

Featured Insight

“This is what using the after the fact AI verification technique is like; if you have to go through this whole verification process after writing your brief, maybe you would have been better off using the old fashioned research tools instead.”

Resources & Deliverables

• Sullivan & Cromwell apology letter with Schedule A of disclosed citation errors: https://lawyeraitoolkit.com/deliverables

About the Show

AI Tools for Practicing Lawyers delivers practical, no-nonsense guidance to help attorneys use AI safely, effectively, and profitably in the modern practice of law.