In depth: AI slop - law's quiet epidemic

‘The AI hallucinations are becoming a nightmare,’ says the owner of a well-established family law firm. ‘We’re up against a litigant in person who has this list of citations from ChatGPT that are all rubbish. It wastes so much time wading through them.’ 

While lawyers presenting false case citations to court have made headlines, a quiet epidemic is making its presence felt in the profession, as unrepresented parties rely on generative AI without realising it has churned out – to use the vernacular – slop.

The problem is certainly not confined to the family courts, and there is no denying the potential access to justice benefits from parties using AI, but there are acute issues related to disputes over children. These are cases where parties are often unable to afford a lawyer and are forced to go it alone. A quick glance at Facebook divorce groups shows how often people are advised to use AI to self-represent and save costs. But there is little understanding that search results cannot always be trusted.

As one family lawyer told the Gazette: ‘There are now three parties to every case. Client, opposition and AI. AI sits in the middle.’

Richard Dawson, an AI strategic counsel to law firm leaders, said the problem is ‘rampant’ and quietly fuelling a structural crisis in the courts.

‘The paradox is that AI is being used to help clear the courts backlog, while at the front door, there is a flood of LiP AI-generated content,’ Dawson said. ‘At 9am on any given Monday, a district judge can review a family court list accelerated by Ministry of Justice backend pilots. By 9:04, she is handed a beautifully fluent, flawlessly formatted 17-page position statement filed by a litigant in person. The prose is confident, but the cited case law is complete fiction.

‘A badly governed public large-language model never says to an unrepresented party, “This claim has no merit”. It never tells them they are wrong; it simply tells them they are right, formats their position elegantly, and drops them into a gridlocked court queue.’

Lawyers are becoming attuned to spotting where parties have relied on AI and are starting to work out appropriate responses.

'AI-generated material can appear sophisticated and persuasive, but be fundamentally flawed'

Paul Maddock and Shaun Johnson, DWF

David Robinson, co-owner of boutique firm Wildcat Law, said the family team has frequently come across AI hallucinations from LiPs and even other solicitors.‘Some of the common issues: long, waffly emails that lack any real point in law. Correspondence clearly applying US legal principles and even procedures. Incorrect details and evidence appearing in correspondence.’

In an article published this month, partner Paul Maddock and associate Shaun Johnson from international firm DWF said AI is driving a rise in unmeritorious LiP claims, creating what has been dubbed ‘legal spam’. As the Gazette reported, the problem was also highlighted at London International Disputes Week earlier this month.

‘[AI] can help individuals understand legal concepts, structure arguments and engage more effectively with the court process,’ they said. ‘The difficulty is that accessibility and reliability do not always align. AI-generated material can appear sophisticated and persuasive, but be fundamentally flawed.’

They said the key challenge for lawyers is to respond to litigants ‘robustly’ while treating them fairly and proportionately. This balanced approach should include maintaining a ‘clear and accessible tone’ in correspondence, recognising the LiP’s position while addressing the deficiencies in their case.

‘Fair treatment does not require tolerance of abusive or meritless litigation, but it does require careful handling,’ they added.

Of course, AI is not just being relied on by unrepresented parties. Lawyers report increasingly common instances of clients thinking they reduce costs by doing much of the legal work without them and effectively turning to a professional when they need a letterhead on top of correspondence.

But dealing with a litigant in person armed with a stack of knowledge but no idea how to test it is a major challenge that is only going to get worse for now.

Sir Geoffrey Vos, master of the rolls, said in a speech earlier this year that the courts face an ‘AI revolution’ due to this increased access.

‘The first port of call used to be a lawyer if one was available and affordable,’ he said. ‘Now the first port of call is ChatGPT or CoPilot.

‘Whatever answer generative AI gives, the would-be litigant in person can easily use it to transform a mass of documents and personal information into an arguable legal claim. This means that, in future, we will see many more civil claims because AI can create them free of charge, where previously claim numbers were limited by the availability and cost of lawyers.’

His proposed solution was online dispute resolution. But the tide of slop may be rising more quickly than the MR’s vision is becoming reality.