Back to News
PolicyAI Understanding briefing

Former lord chancellor urges costs regime to curb AI‑generated legal slop

Sir Robert Buckland warned that AI‑generated filings risk flooding UK courts with unnecessary material and suggested judges use existing costs rules to penalise irresponsible use by litigants in person.

4 min readRead the linked source
Source-provided image accompanying Former lord chancellor urges costs regime to curb AI‑generated legal slop
Source referenceSource recorded
Publisher
lawgazette.co.uk
Source link
lawgazette.co.ukhttps://www.lawgazette.co.uk/news/use-costs-to-clamp-down-on-ai-slop-urges-former-lord-chancellor/5127998.article
Source type
Linked source — primary-source status has not been established.
ContextUnderstand this in 60 seconds

Start here

Test yourselfAI Ethics Quiz

What happened

Sir Robert Buckland KC, former lord chancellor, told the Costs Law Reports conference that the rise of AI‑generated legal documents is creating a new costs problem for the courts. He urged judges to apply the existing CPR 44 costs regime to discourage litigants in person from flooding proceedings with “AI slop” – large, superficially impressive but unnecessary pleadings, statements and arguments produced by AI tools. Buckland said the court should make clear at the outset that AI can be used as an aid, but the filing party remains responsible for accuracy, relevance and proportionality, and that unreasonable costs arising from AI‑generated material may be taken into account when assessing costs orders.

At the Costs Law Reports conference, Sir Robert Buckland KC warned that AI has caused a "profound change in the economics of litigation" by allowing litigants in person (LiPs) to generate large volumes of legal material at virtually no marginal cost.

He described "AI slop" as the production of superficially impressive but unnecessary documents—pleadings, witness statements, skeleton arguments, and lists of authorities—created by AI tools such as ChatGPT, Claude or Gemini.

Buckland argued that while existing CPR 44 powers already allow courts to make costs orders for unreasonable conduct, the rules should be explicitly applied to AI‑generated conduct, making clear that the filing party remains responsible for accuracy and relevance.

He cautioned against a blanket "LIP + AI = costs sanction" approach, emphasizing the need to distinguish between uninformed use and deliberate abuse, and suggested that judges issue early notices outlining AI‑use expectations and potential cost consequences.

Source details: lawgazette.co.uk ↗

Why it matters

If courts adopt Buckland’s suggestion, AI‑driven over‑production could be curbed, protecting taxpayers and preserving the efficiency of the justice system. The proposal highlights a concrete risk that AI lowers the marginal cost of drafting extensive legal submissions, potentially encouraging litigants in person—who often lack legal training—to submit voluminous, low‑quality documents. By tying costs sanctions to AI‑generated conduct, the judiciary could preserve proportionality, deter abuse, and maintain access to justice for those who cannot afford professional representation. The guidance also signals to legal practitioners that reliance on AI does not absolve them of responsibility, reinforcing professional standards in an era of rapidly advancing generative tools.

The suggestion directly addresses a growing operational burden on UK courts, where AI can enable self‑representing parties to submit overly lengthy or irrelevant filings, increasing costs for opponents and the public purse.

By linking costs sanctions to AI‑generated material, the proposal aims to preserve proportionality in litigation, a core principle of civil procedure, while still allowing LiPs to access AI assistance for genuine assistance.

The guidance reinforces professional responsibility, reminding lawyers that reliance on AI does not provide a defence against inaccurate or frivolous submissions, thereby upholding standards of legal practice.

If adopted, the approach could set a precedent for other common‑law jurisdictions grappling with similar AI‑driven challenges in their courts.

Interactive Mechanism

Interactive Mechanism: How It Actually Works

Explore the underlying technology behind this development interactively.

Model Parameter Size:8B Parameters
VRAM Required5.5 GBGPU memory footprint
Target HardwareMacBook / Single GPUDeployment tier
Privacy100% Air-GappedLocal device capability
Core takeaway: Small, quantized models (3B–8B) now run directly inside smartphones and laptops with complete data privacy, while mammoth 400B+ models remain the domain of datacenter clusters.
Interactive Concept Check+10 Points
AI Ethics Quiz

Impossibility results in algorithmic fairness (e.g. Kleinberg et al., Chouldechova) show what?

What to watch next

Watch for any formal guidance issued by the Civil Procedure Rules Committee or the Ministry of Justice that codifies Buckland’s recommendations. Courts may begin issuing pre‑litigation notices that outline AI‑use expectations, and future case law could clarify how costs judges assess AI‑related excess. Monitoring how litigants in person respond to any new costs sanctions will indicate whether the approach effectively reduces frivolous filings without unduly penalising genuine self‑representants.

Potential issuance of formal practice directions or amendments to CPR 44 that explicitly reference AI‑generated conduct.

Judicial decisions that apply costs sanctions to AI‑generated filings, which would clarify the practical enforcement of Buckland’s recommendations.

Reactions from legal professional bodies, such as the Law Society and the Bar Council, regarding the balance between access to justice for LiPs and the need to curb abusive AI use.

Data on whether the introduction of AI‑related costs orders leads to a measurable reduction in excessive filings without deterring legitimate self‑representation.

Related guides & quizzes

AI EthicsFuture of AIWhat is AI?Test what you know — try a free AI quizLook up an AI term in our glossaryFollow the AI regulation tracker
Found this useful?