In partnership with
This is the monthly catch-up on AI developments from March 2026 that UK lawyers may have missed. The weekly issues covered the main stories; this mop up surfaces context and items that fell between the cracks. It should take about five minutes to skim.
Last month in UK legal AI
March was dominated by two common themes: copyright and privilege.
The month opened with the Upper Tribunal judgment in Munir, which held that uploading confidential documents to ChatGPT waives legal professional privilege. Simmons & Simmons published a practical guide and policy framework almost immediately, and the story set the tone for how firms thought about AI risk for the rest of the month (6 March issue).
On copyright, the House of Lords Communications and Digital Committee published its 180-page report calling for a licensing-first approach, rejecting the government's preferred opt-out model (12 March issue). Days later the government confirmed it had abandoned the opt-out, after 88% of consultation respondents backed mandatory licensing (20 March issue).
Beyond those two threads, regulators were active across the board. The CMA published guidance on agentic AI and consumer law, the CJC consulted on mandatory AI declarations in witness statements, and Irwin Mitchell published research showing that over a third of in-house legal teams have seen a rise in AI-generated claims (27 March issue). Anthropic launched its Claude Partner Network, opened up agentic desktop capabilities, and continued to reshape the competitive landscape between foundation model providers and legal tech incumbents. And Harvey announced its integration with Microsoft 365 Copilot, due in Q2 2026.
The overall picture is one of rapid movement on multiple fronts: regulatory, judicial, and commercial. If you only read one back issue, make it the 20 March edition, which covered the government copyright report, the CMA agentic AI guidance, and the Anthropic Partner Network in a single week.
Stories you may have missed
DRCF publishes joint foresight paper on agentic AI: On 31 March, the Digital Regulation Cooperation Forum (CMA, FCA, ICO, and Ofcom) published "The Future of Agentic AI," mapping a five-level autonomy spectrum and cataloguing risks including algorithmic collusion, prompt injection, and consumer rights challenges. The paper affirms that existing UK regulatory frameworks apply to agentic systems and commits to further horizon-scanning work in 2026 and 2027. For lawyers advising on consumer-facing AI or financial services, this is the regulatory baseline for agentic systems in the UK. (DRCF / Lewis Silkin)
AI-first conveyancing firm Keith raises £2m: Keith, a new UK law firm planning to automate roughly 80% of the residential conveyancing process, raised £2 million in seed funding and aims to launch in Q3 2026. The client-facing layer is a 24/7 AI agent accessible by phone and WhatsApp. Critical steps, including fund transfers, remain human-handled. Keith is seeking authorisation from the Council for Licensed Conveyancers rather than the SRA. Whether or not Keith succeeds, the model is a signal of where investor money is going and how new entrants are thinking about legal service delivery. (Artificial Lawyer / Legal Futures)
Government launches "Growing up in the online world" consultation with AI obligations: On 2 March, DSIT launched a consultation on children's online safety that includes proposals for new obligations on AI chatbots and generative AI services, stricter age assurance, and a possible statutory minimum age for social media. The consultation closes on 26 May 2026. On 12 March, Ofcom and the ICO followed up with coordinated letters to platforms demanding effective age checks. Lawyers advising technology clients or education providers should have this on their radar. (GOV.UK)
ICO flags AI risks in automated recruitment: On 31 March, the ICO published guidance warning that employers using AI in recruitment must ensure proper safeguards are in place. The guidance focuses on transparency, fairness, and the right of job applicants to understand how automated decisions are made about them. Employment and data protection teams should note this, particularly where clients are deploying AI screening tools. (ICO)
GPT-5.4 ships with improved legal accuracy: OpenAI released GPT-5.4 on 5 March, claiming a 33% reduction in false claims compared to GPT-5.2. Harvey reported a 91% score on its BigLaw Bench for document-heavy legal work, and a separate LegalOn review found a 21% improvement in contract drafting accuracy. The weekly issues touched on this only briefly. For firms evaluating which model to use for legal work, the benchmarks are worth reviewing. (Integrated Cognition / Artificial Lawyer)
HMCTS J-AI listing pilot confirmed in written answer: On 9 March, the government confirmed in a written parliamentary answer that HMCTS is trialling J-AI, an AI-assisted listing tool for judges, in Preston and Isleworth Crown Courts. The tool flags cases that may need court action to progress and assists with scheduling. Evaluation criteria include reductions in administrative effort, improved timeliness, and listing accuracy. If successful, national roll-out is planned. (TheyWorkForYou, Previous HMCTS Blog)
Artificial Lawyer: "Prompts are a crutch, legal AI needs memory": A piece from Artificial Lawyer argued that prompt-based legal AI tools become outdated and inconsistent over time, and that the next step is systems with memory that learn from past decisions and improve continuously. The argument is that context retention, not better prompts, is what will make legal AI reliable at scale. Worth reading alongside the Electra Japonas piece on context that the weekly issue covered on 27 March. (Artificial Lawyer)
Legal IT Insider: the future of client relationships in the GenAI era: Published on 27 March, this piece examines how generative AI is changing law firm business development and client management, with firms increasingly using data to understand client needs and target their efforts. For partners and BD teams, this is a useful prompt to think about whether your firm's approach to client relationships is keeping pace. (Legal IT Insider)
Ad Break
1,000+ Proven ChatGPT Prompts That Help You Work 10X Faster
ChatGPT is insanely powerful.
But most people waste 90% of its potential by using it like Google.
These 1,000+ proven ChatGPT prompts fix that and help you work 10X faster.
Sign up for Superhuman AI and get:
1,000+ ready-to-use prompts to solve problems in minutes instead of hours—tested & used by 1M+ professionals
Superhuman AI newsletter (3 min daily) so you keep learning new AI tools & tutorials to stay ahead in your career—the prompts are just the beginning
How did we do?
Reply and tell me what you would like included in future mop ups.
We read every email!
UK Legal AI Brief
Disclaimer
Guidance and news only. Not legal advice. Test outputs and apply professional judgment.