There is no shortage of legal-AI news; there is a shortage of legal-AI news that matters to a firm's actual obligations. Most of it is product launches and funding rounds. Very little of it changes what a COLP has to do on Monday.
This is a running brief on the developments that do — regulation, court decisions, and the underlying data — for UK and EU firms. It's updated regularly, every item is sourced, and the framing is what it means for you, not who raised what. Newest developments first.
Last updated: 2 July 2026.
Regulation
EU AI Act high-risk deadline deferred to December 2027. The EU's Digital Omnibus package, approved by the European Parliament in June 2026, pushed the Annex III high-risk obligations from 2 August 2026 to 2 December 2027, and moved Article 50 transparency duties to December 2026. Why it matters: the "August 2026 cliff" much of the market braced for has softened for high-risk uses — but the prohibitions and AI-literacy duty (Article 4) have been binding since February 2025 and are unaffected. Don't read the deferral as "nothing until 2027." Full detail in our EU AI Act summary.
SRA guidance confirms the COLP owns AI compliance. The SRA's compliance guidance on AI and technology sets out that it expects, as a minimum, the COLP to be responsible for regulatory compliance when new technology is introduced, with governance, systems and controls around its use. Why it matters: there is no separate AI rulebook — AI falls under the existing Codes, and accountability is already assigned. See SRA compliance for AI.
ICO guidance frames AI as a data-protection question. The ICO's Guidance on AI and Data Protection organises AI obligations around UK GDPR principles — accountability, transparency, lawfulness, accuracy, security, and data minimisation. Why it matters: for firms handling personal and special-category data, AI use is a data-protection compliance issue as much as an SRA one.
The courts
AI hallucination cases pass 1,668 — 653 involving lawyers. Damien Charlotin's public database of court decisions addressing AI-fabricated content reached 1,668 cases as of 2 July 2026, with 653 involving practising lawyers rather than self-represented litigants. Why it matters: the risk is no longer theoretical or confined to amateurs; it's a documented, accelerating pattern reaching regulated professionals. Our hallucination cases tracker covers the detail.
UK courts escalate from warnings to referrals. Following Ayinde v Haringey and Al-Haroun v Qatar National Bank ([2025] EWHC 1383 (Admin)), where fabricated citations drew wasted-costs orders and regulator referrals, a 2026 UK case saw a judge refer solicitors to the SRA over AI-generated citations, holding that admonishment alone was insufficient. Why it matters: the leniency that came with novelty is gone. Verifying AI output against source is now treated as a baseline professional duty. See how to verify AI legal citations.
The data
61% of UK lawyers now use generative AI — but governance lags. LexisNexis's The AI Culture Clash survey found 61% of UK lawyers using generative AI day-to-day, up from 46% in January 2025, while two-thirds described their firm's AI culture as slow or non-existent. Why it matters: adoption has outpaced governance — the gap where firms are most exposed.
Only a minority of organisations have an AI strategy. Thomson Reuters' Future of Professionals 2025 report found only 22% of organisations have a visible AI strategy, though those that do are markedly more likely to see a return. Why it matters: the differentiator between firms isn't whether they use AI — nearly all do — but whether they govern it deliberately.
Compliance officers are under-supported. The SRA's December 2025 thematic review found only one of thirty-six compliance officers could describe all the material requirements of their role, and reporting to the SRA was strikingly low. Why it matters: the people accountable for governing AI are stretched before AI is added to their remit.
What we're watching
- The EU AI Act's remaining phase-ins — GPAI enforcement in practice, and the run-up to the December 2027 high-risk date.
- UK AI regulation — whether the current framework-plus-guidance approach shifts, and the outcome of sector consultations.
- Further UK enforcement on AI-generated citations, and whether the SRA issues more targeted AI guidance.
- Adoption vs governance data — whether the gap between the two narrows through 2026.
How to use this brief
If you're a COLP or managing partner, the throughline across all of it is the same: AI adoption is mainstream and accelerating, the obligations are already live, and the exposure sits in the gap between using AI and being able to prove you governed it. The practical response is a governance framework that produces evidence — not another policy document.
Want this brief in your inbox, and the governance layer that turns it into action? LegalAI Space builds AI for legal teams with verification and an audit trail so you're ready for what's live today. Book a 30-minute call with Daman.
Related reading
- SRA compliance for AI: the complete guide — the obligations behind the headlines.
- The EU AI Act, summarised — risk tiers, dates, and who's affected.
- 1,600+ AI hallucination cases — the court record, tracked.