How to Verify Citations and Catch AI Hallucinations—Webinar

Ensuring your legal briefs are entirely accurate is more stressful than ever with the rise of AI-assisted legal drafting.

In a recent webinar with Today’s Managing Partner, Ross Guberman, Founder and CEO of BriefCatch, discussed how the challenge of verification has evolved and how legal teams can navigate misunderstandings regarding AI hallucinations and errors in 2026.

The session, “Before You File: Verifying Citations or Hallucinations in AI-Assisted Legal Drafting,” emphasized the importance of recognizing how the legal tech landscape has shifted. It called for an understanding of how large language models (LLMs) have changed, altering how professionals must approach document review before submitting briefs to prominent courts. The webinar also demoed BriefCatch’s new “RealityCheck” as one tool for cross-referencing legal citations against primary sources to instantly flag inaccurate quotes, misstated holdings, and hallucinated authorities.

Guberman noted that while the classic stories of completely fabricated cases resulting in court sanctions used to dominate the news, the technology has entered a new phase.

“What’s happening is they’re citing cases that exist, but what they’re saying about the cases is often very far afield,” he said. “And that’s kind of like the new hallucination problem that you’re seeing more and more of.”

Key takeaways from the session:

• Understand the new hallucination problem: Legal professionals can no longer just scan a table of authorities to ensure a case exists. Modern LLMs like ChatGPT and Claude have improved at not making up cases wholesale, but they frequently struggle to understand holdings the way a litigator does, leading to misrepresented legal principles.
• Integrate verification into your daily workflow: Attorneys do edit, proofread, and site-check in separate silos—these tasks naturally meld together. To save valuable time and maintain focus, teams should utilize tools that operate directly within Microsoft Word rather than constantly switching between different browser tabs and external LLM portals.
• Look beyond AI-specific errors: The broader challenge of document integrity and citation accuracy has existed since long before generative AI. Effective verification platforms should not only flag fake quotes or hallucinated authorities, but also assist diligent lawyers with older, subtle issues like missing words in copy-pasted text, wrong procedural postures, or misleading citation signals.

Ultimately, protecting your firm from court sanctions requires a more deliberate and modern approach to document review. Watch the full recording to learn how to shift from basic fact-checking to advanced AI verification.

Read more about legal writing in Ross Guberman’s Today’s Managing Partner columns here.