In partnership with
The legal AI market has been building towards something for months. This week, OpenAI made a move by making its legal ambition concrete and hiring the founder of Ironclad, joining Anthropic and Microsoft in a three-way competition for law firm business. Meanwhile, the economics of legal AI are shifting in ways that will affect every firm's budget.
OpenAI hires Ironclad founder to lead its legal vertical launch
Jason Boehmig, founder of contract lifecycle management company Ironclad (valued at $3.2 billion), joined OpenAI on 1 June to "lead product for the legal vertical." His LinkedIn now reads "Building AGI for law at OpenAI."
The hire confirms what Artificial Lawyer first reported on 18 May: OpenAI is building a legal-specific offering, likely branded as "Codex for Legal," structured as plugins and tools within its existing Codex platform. Boehmig is not a tech tourist. He left a career as a law firm associate to build Ironclad from scratch, growing it to over 700 staff and hundreds of millions in recurring revenue before stepping down as CEO in 2025. He remains Ironclad's executive chairman.
This matters because it completes a set. Anthropic has been building out Claude for Legal since February (covered in this newsletter on 15 and 22 May). Microsoft released its Legal Agent for Word in April. OpenAI is the third major foundation model provider to target law firms directly, and unlike the other two, it also has a "Deployment Company" rolling out forward-deployed engineers to help enterprise customers implement AI at scale.
For UK firms, the practical consequence is clear: the range of viable legal AI providers just expanded again, and the competitive pressure between them should drive both better products and lower prices. But it also means the market is moving faster than many firms' procurement processes. Firms that have not yet begun a structured evaluation of these platforms risk falling behind, not because AI is essential today, but because the integrations, data practices, and vendor relationships being established now will shape how legal work is delivered for years.
Takeaways
Act: If your firm is still in the "wait and see" phase on legal AI providers, this is a good moment to begin a structured comparison of the three Big Tech offerings (Claude for Legal, Microsoft Legal Agent, and the forthcoming OpenAI product) alongside your existing legal tech stack. You do not need to commit yet, but you do need to understand the options.
Watch: The shape of OpenAI's legal product when it launches. Will it require engineering support (like Codex), or will it be accessible to lawyers without technical assistance? The answer will determine how relevant it is for mid-market UK firms.
Risk: Going all-in with one provider locks you into that provider's models. Established legal tech vendors can choose whichever LLMs to work from as models evolve, which some would argue gives them an edge. Consider whether flexibility matters more to your firm than deep integration with one platform.
Read: Artificial Lawyer | LawSites | Legal IT Insider
On your radar
Claude for Legal now has over 90 customisable AI agents on GitHub: Anthropic has quietly built out a catalogue of named workflow agents, each handling a specific legal task end to end (Vendor Agreement Reviewer, DSAR Responder, Termination Reviewer, Claim Chart Builder, and dozens more). Each agent can be modified in plain language, without engineering skills, and some are designed to run continuously on incoming documents or emails. Mark Pike, associate general counsel at Anthropic, emphasised that the tooling is "designed to make that review easier, never to skip it." The latest Opus 4.8 model also improves uncertainty flagging, which is welcome given the hallucination problems the profession continues to grapple with. Why it matters for UK lawyers: The granularity is where these tools become useful. A generic "contract review" tool is limited; a named agent that sweeps signed agreements weekly for playbook deviations is a different proposition. If your firm is evaluating Claude for Legal, browse the GitHub catalogue before your next vendor call. (Artificial Lawyer)
Kirkland & Ellis hints at fine-tuning LLMs for its own legal AI model: The world's highest-grossing law firm is committing $500 million over the next three to four years (starting with $100 million in 2026) to build a custom AI platform, and recent job listings suggest the project may involve fine-tuning open-source LLMs on the firm's own data using on-premise GPU infrastructure. Around 180 staff will work on the project, with input from 250 of the firm's lawyers, including 100 partners. The tools will not be available to other firms. Why it matters for UK lawyers: Most UK firms will not have $500 million to spend. But the strategic question Kirkland is answering is one every firm should ask: do you build, buy, or combine? If the largest firms start training models on their own data and gaining measurable performance advantages, mid-market firms may need to consider how they access similar capabilities through vendors or consortia. Discuss with your innovation committee whether your current approach accounts for this possibility. (Artificial Lawyer | Legal Cheek)
Legal AI token costs are rising as subsidised pricing ends: Anthropic has shifted Claude Enterprise to a hybrid pricing model that charges clients directly for token usage on top of a reduced flat fee, and the broader trend is the same across providers. The tasks lawyers now perform with frontier models (reasoning, agentic workflows, large document processing) are far more token-intensive than the chatbot-style queries of two years ago, and the era of flat-rate subsidised access appears to be ending. Why it matters for UK lawyers: If your firm adopted AI tools under flat-rate pricing, check whether your current contract includes token-based escalation clauses or usage caps. Ask your vendor what a typical month's token consumption looks like for your use patterns, and build that into next year's budget. Firms that do not track usage now risk an unpleasant surprise at renewal. (Artificial Lawyer | Law.com)
One in five US federal court filings now contains AI-generated text: A study by researchers at MIT and the University of Southern California, published this week, analysed 4.5 million civil lawsuits and found that the share of federal complaints flagged as containing AI-generated writing rose from 1% in 2023 to 18% in early 2026. Pro se (self-represented) filings have nearly doubled since ChatGPT launched, with 41,490 pro se filings in fiscal year 2025 alone. The increase is overwhelming some courts. Why it matters for UK lawyers: The UK hallucination tracker already logs 64 documented cases of AI-generated false authorities reaching English courts, and the trend in self-represented litigant filings is likely to follow a similar pattern here. For litigators, the practical consequence is that AI-generated material from the other side (or from litigants in person) will increasingly need to be verified, not just your own. Raise this with your litigation team as part of any case management discussion involving unrepresented parties. (MIT Technology Review)
Ad Break
In order to help cover the running costs of this newsletter, please check out the advert below. In line with my promises from the start, adverts will always be declared.
Your prompts are leaving out 80% of what you're thinking.
When you type a prompt, you summarize. When you speak one, you explain. Wispr Flow captures your full reasoning — constraints, edge cases, examples, tone — and turns it into clean, structured text you paste into ChatGPT, Claude, or any AI tool. The difference shows up immediately. More context in, fewer follow-ups out.
89% of messages sent with zero edits. Used by teams at OpenAI, Vercel, and Clay. Try Wispr Flow free — works on Mac, Windows, and iPhone.
For Review
"MCP: The Standard that Decides Legal AI's Future" (Artificial Lawyer)
An overview of why the Model Context Protocol matters for legal technology. MCP, originated by Anthropic and now governed by the Linux Foundation's Agentic AI Foundation (whose members include OpenAI, Google, Microsoft, and AWS), is becoming the common standard for connecting AI tools to external systems. For legal, this means AI agents can connect to case law databases, document management systems, and citation validators through a single protocol rather than bespoke integrations. The piece notes that iManage launched MCP support on 14 May and NetDocuments is moving in the same direction. If your firm is making infrastructure decisions about legal AI, understanding MCP is worth the ten minutes.
Read or listen: Artificial Lawyer
"How courts are coping with a flood of AI-generated lawsuits" (MIT Technology Review)
The full MIT Technology Review treatment of the MIT/USC study mentioned above. The piece examines not just the numbers but the practical consequences for courts: judges spending more time filtering low-quality filings, clerks dealing with fabricated citations, and a growing tension between making courts accessible to self-represented litigants and managing the volume of AI-generated material. The US context is different from the UK, but the underlying dynamic (AI lowering the cost of producing legal documents while not improving their quality) applies here too.
Read or listen: MIT Technology Review
"OpenAI Targets the Legal Vertical: What Happens to Legal Tech?" (Artificial Lawyer)
A companion piece to the Boehmig hire story, exploring what happens to the legal tech market when all three major AI providers offer their own legal-specific products. The analysis considers two scenarios: one where Big Tech dominates and absorbs much of the market (particularly in-house legal), and another where the providers lose interest or under-invest, leaving the market competitive and diverse. Worth reading for the strategic framing if your firm is in the middle of a vendor evaluation.
Read or listen: Artificial Lawyer
Practice Prompt
Try the below prompt to evaluate which legal AI provider fits your firm. Ensure you fill in context and constraints and other aspects marked with {}. Remember to adhere to the Golden Rules and do not upload confidential or privileged information to public tools.
You are a legal technology evaluation assistant. Your task is to help a UK law firm compare legal AI providers and produce a structured assessment it can use internally.
Firm details:
- Firm size and type: {e.g., "6-partner high street practice" / "50-lawyer regional firm" / "200-lawyer City firm"}
- Practice areas: {e.g., "commercial litigation, corporate, employment" / "personal injury, clinical negligence" / "private client, residential conveyancing"}
- Current technology stack: {e.g., "iManage for DMS, PMS is Proclaim, Microsoft 365" / "NetDocuments, Elite 3E, Microsoft 365" / "LEAP, Google Workspace"}
- AI tools currently in use (if any): {e.g., "Microsoft Copilot under enterprise licence" / "CoCounsel for document review" / "none formally adopted"}
- Budget range for AI tools: {e.g., "under £5,000/year" / "£20,000-£50,000/year" / "no fixed budget, ROI-driven"}
- Technical capacity: {e.g., "no dedicated IT staff" / "small IT team, no AI expertise" / "innovation team with some engineering capability"}
Using the three major provider offerings now available (Claude for Legal from Anthropic, Microsoft Legal Agent, and the forthcoming OpenAI legal vertical) plus established legal tech vendors (e.g., Harvey, CoCounsel, Luminance, Lexis+ AI), produce:
1. **Provider mapping**
For each provider relevant to this firm's size and practice areas, assess:
- What it offers today (not what is announced or planned)
- How it integrates with this firm's existing technology stack
- Pricing model (flat fee, per-seat, token-based, hybrid) and likely cost at this firm's scale
- Whether it requires technical support to deploy or can be used by lawyers directly
- LLM flexibility (single model vs. choice of models)
- Data handling: where does client data go, and what are the contractual commitments on confidentiality and training data exclusion?
2. **Workflow fit**
Identify the 3-5 workflows at this firm where AI would deliver the clearest return (e.g., contract review, legal research, document drafting, due diligence, case summarisation). For each workflow, note which provider(s) could handle it today and what gaps remain.
3. **Risk assessment**
For each shortlisted provider:
- Regulatory risk: does using this tool create any SRA compliance exposure (confidentiality, supervision, competence)?
- Vendor lock-in risk: how difficult would it be to switch providers after 12 months of use?
- Cost risk: is pricing likely to increase (e.g., token-based models where usage grows with adoption)?
4. **Recommendation**
Based on this firm's size, budget, technical capacity, and practice areas, recommend:
- A primary provider to evaluate first (with reasoning)
- A secondary option to compare against
- Any providers that are not a realistic fit for this firm and why
Constraints:
- {Add any firm-specific constraints, e.g., "We cannot use any tool that stores data outside the UK" / "Partners will not use any tool that requires a command line" / "We need something live within 60 days"}
- Apply the regulatory framework of England and Wales throughout
- Do not invent product features or pricing. Where information is uncertain, say so
- This is an evaluation framework, not legal advice or a product endorsement
How did we do?
Hit reply and tell me what you would like covered in future issues or any feedback. We read every email!
Thanks for reading,
Serhan, UK Legal AI Brief
Disclaimer
Guidance and news only. Not legal advice. Always use AI tools safely.
Recommended Newsletters
Below are a few newsletters that I recommend, for various reasons. Check them out!
