BriefCatch Targets AI Errors With Expanded Legal Writing Suite

Generative AI can speed up legal research, drafting, and revision, but it can also introduce fabricated citations, inaccurate quotations, misstated court holdings, and unsupported legal claims.

The new self-serve online application from BriefCatch lets individual attorneys, small law firms, law students, and courts purchase direct access to its legal writing and citation verification tools.

Last month, BriefCatch combined writing, verification, and training tools in a unified suite designed to help legal professionals review their work before it reaches a client, court, or opposing counsel.

Attorneys can also use BriefCatch to flag errors in opponents’ briefs, including unsupported claims that could mislead the court.

Clear, accurate legal writing should not require a tradeoff between speed and accuracy, according to Ross Guberman, founder and CEO of BriefCatch.

“Nor should it be something only the largest firms can access,” he said. “With the new BriefCatch, we’re giving all legal professionals the tools to produce stronger, more accurate work today, while becoming better writers and reviewers over time.”

Writing, Verification, and Training in One App

BriefCatch’s new online app offers direct, self-serve purchasing for all legal professionals. Guberman expects individual lawyers’ use of the app to encourage adoption across larger firms.

He said lawyers shouldn’t need to navigate a large firm’s bureaucracy to get a second set of eyes on a brief. The new online app makes it easier to try BriefCatch, subscribe, and get back to work.

For years before creating BriefCatch, Guberman taught writing techniques to lawyers and judges who would leave a workshop eager to try them. But then a deadline would intervene, or they could not remember an example that had impressed them.

“When I decided to dip my toes in legal tech, I wanted to put that guidance beside them when and where they needed it. Years later, our new app puts editing, verification, and training in one place, which especially helps lawyers who don’t have a colleague down the hall willing to review a draft,” he told TechNewsWorld.

He sees individual enthusiasm spearheading firm-wide adoption. That also occurred with his existing application, RealityCheck, which detects fabricated legal citations, misstated holdings, and AI hallucinations in legal briefs before they are filed.

“We’ve seen that more than once. One lawyer gets excited about BriefCatch or RealityCheck, tells colleagues, and becomes its champion inside the firm. Self-serve makes that first step easier, but it doesn’t replace the security reviews and administrative controls that firms still need,” he said.

Guberman added that RealityCheck offers other opportunities, including in doctrinal classes. It teaches how to take a flagged claim, read the opinion, and defend a conclusion — including a conclusion that the tool is wrong.

“More than ever, students need to learn that a source, a summary of that source, and an argument built on it are three different animals,” he explained.

How Detection Works

BriefCatch checks citations and legal claims for errors, regardless of whether a document was drafted by a person or an AI tool.

Guberman noted that an unsupported proposition needs better support, whether it came from a chatbot or a seasoned litigator. It doesn’t matter whether it was written in 2026 or 1976.

“Lawyers were misquoting cases and overstating holdings long before anyone had heard of generative AI. What AI changes is how persuasive a mistake can look,” he said.

The software checks whether the sources support an argument, even when polished prose and well-formatted citations make it appear sound.

“We’re also vigilant about subtler errors involving, for example, facts that a court merely mentions as opposed to facts that are integral to the holding,” said Guberman about the new software’s added reliability.

Litigators Scrutinize AI Errors

Andrew Adams, partner and chief administrative officer at DarrowEverett, views the BriefCatch upgrade as an essential innovation. Automated error-checking on an opponent’s brief, scanning for unsupported propositions or altered quotes, is becoming increasingly important in day-to-day litigation practice.

He agreed that hallucinations and shadow AI — employees using unapproved AI tools — are two serious risks facing the legal industry today. Traditionally, brief review meant checking whether a case was still good law and relevant to the argument. Today, litigators also run opposing briefs through citation verification to hunt for AI hallucinations and misstated quotes.

“Given the market-wide promulgation of AI in legal, this type of review and check has become part of standard practice in litigation,” he told TechNewsWorld.

Adams noted that traditional drafting errors have always existed. Today’s practitioners are far more concerned with entirely fabricated citations propping up an adversary’s core argument.

“We’ve already adopted various versions of these tools in our day-to-day practice. I think we are going to see market-wide adoption of these tools in the majority of law firms,” he said.

He believes these checks will become the market standard in the near future, with every attorney reviewing both their own filings and those of opposing counsel.

Human Review Remains Essential

From Adams’s perspective, as AI becomes more ingrained in daily practice across the legal spectrum, it becomes increasingly important for legal workers to use review technology to evaluate substantive accuracy. This is especially true where attorneys handle a large volume of drafting work, not only in private practice but in the public sector as well.

“We have all witnessed the increase in material misstatements, hallucinated cases, and drafting errors when general AI is utilized in drafting substantive briefs,” he said.

Adams hopes the combination of system-level prompting through memory and/or backend instructions, citation checking, verification technology, and final human review will reduce these types of errors in the future. He cautioned that the most essential part of the review will always be human oversight of these tools.

“In short, it is always best practice, and in the case of legal, our ethical obligation, to have final human review and sign-off before any filing or any other external communication related to a generated document,” he said.

Newer AI Models Still Make Legal Errors

Adams sees fewer fabricated cases with newer general-purpose AI models when users provide more detailed instructions, particularly through system-level prompting. His bigger concern is subtly misstated holdings.

“Generally, what we encounter the most are those slightly misstated quotations rather than misstated holdings when utilizing legal AI tools,” he offered.

He suggested that dedicated verification layers help catch some of those issues. Tools from newer vendors, or drafting add-ins from Westlaw, Lexis, or other legal databases that tie directly into Microsoft Word, can link back to cases and citations and perform citation checks.

“When designing your policies and best practices regarding agentic workflows, users should always plan where human oversight and decision-making comes into play, as in the human-in-the-loop,” he said.

RealityCheck Flags Misleading Uses of Case Law

Guberman explained that RealityCheck performs checks such as whether a quotation matches the cited opinion. Even that comparison can require legal reasoning.

For example, if the brief-writer properly uses ellipses to truncate a quote, is the missing language essentially filler, or does it convey a point the writer had reason to downplay?

Some of the program’s best flags relate to real cases cited in misleading ways. These might include a brief that presents a concurrence as the majority’s holding, attributes a litigant’s argument to the court, or treats a decision allowing a claim to proceed as though the court had reviewed the claim on the merits.

Guberman noted that overstatements deserve scrutiny, too. As examples, he cited using the phrase “all courts” when the support reaches only some, or a broad rule derived from a decision with a crucial exception.

“On the other hand, we realize that lawyers have every right to draw analogies and argue for extensions of existing law. But sometimes those efforts go too far,” he said.

Teaching Legal Writing With BriefCatch

Guberman outlined ways law schools can use BriefCatch to help students develop their writing and review skills.

For example, a professor could assign a draft, have students review BriefCatch’s suggestions, and ask them to explain one edit they accepted and one they rejected. Students could then apply the same technique to a new paragraph without the tool.

“That last step matters. It shows whether they learned something beyond how to operate legal tech,” he explained.

BriefCatch’s Learning Hub includes a 90-module legal writing curriculum that faculty can draw on for assignments on organization, sentence structure, or contract drafting.