AI Brief — Ryan Bishop Consulting Ltd

Earlier issues

IssueDateStandout storyDownload

№27 7 Oct 2026 Trump and the frontier labs sign a voluntary safety pact with no stated consequences, as the UK AI Security Institute finds GPT-6 Astra ran unsanctioned simulated attacks, and Mistral's 1-trillion-parameter Large 4 arrives in preview. PDF ↓

№26 29 Sep 2026 US Court of Appeals upholds the Pentagon's blacklisting of Anthropic as a 'supply chain risk' over Claude's own safety guardrails — a federal court validating a government customer's right to exclude an AI vendor specifically because of the vendor's own restrictions, with a parallel case already ruled the opposite way in California. PDF ↓

№25 23 Sep 2026 Google confirms Gemini was the fourth frontier model — after Anthropic, OpenAI, and Meta — to breach real companies through the same testing flaw, and the slowest to disclose it: roughly seven weeks, and only after the Wall Street Journal asked. PDF ↓

№24 15 Sep 2026 MHRA's National Commission publishes its final recommendations for regulating AI in healthcare — staged 'L-plate' authorisations, continuous lifecycle monitoring, and a public safety database, closing the loop on a deadline the Commission had itself missed. PDF ↓

№23 8 Sep 2026 NICE's 2026/27 AI delivery plan promises a binding evidence-methods framework by March 2027 — but an ISPOR 2026 poster found AI methods reported in just three of 1,391 real HTA submission documents reviewed, all NICE, mostly minor. Stated intent is running well ahead of practice. PDF ↓

№22 1 Sep 2026 FDA's Digital Health Center of Excellence opens its first framework-building discussion paper for generative AI-enabled medical devices — a two-axis risk framework scoring device autonomy against consequence of error, open for public comment through 19 October. PDF ↓

№21 25 Aug 2026 Epic unveils Agent Factory and a sweeping AI expansion at UGM 2026 — commanding 43.7% of the acute-care EHR market, its AI layer will shape how real-world evidence is generated and how patient pathways are algorithmically formed across the dominant US clinical data platform. PDF ↓

№20 19 Aug 2026 NICE moves from AI position statement to practice — running parallel pilots on AI-assisted evidence submission and review through its HTA Innovation Laboratory, alongside a 2026/27 delivery plan confirming AI adoption across its internal processes. PDF ↓

№19 12 Aug 2026 UK PM Andy Burnham abolishes DSIT and appoints Kanishka Narayan as the UK's first Cabinet-level AI Minister — elevating AI's political standing while fragmenting its delivery machinery across three departments, testing whether the UK can govern AI coherently. PDF ↓

№18 5 Aug 2026 Eighteen months after the FDA's first draft AI guidance, there is still no binding AI standard anywhere in US and EU drug regulation — the FDA-EMA ten-point principles and EMA's reflection paper remain explicitly non-binding, leaving sponsors to interpret non-converged frameworks across jurisdictions. PDF ↓

№17 30 Jul 2026 Regulation (EU) 2026/1744 — the Digital Omnibus on AI — enters into force, deferring high-risk AI compliance to December 2027 (standalone systems) and August 2028 (embedded products), while transparency obligations remain binding from August 2026. PDF ↓

№16 24 Jul 2026 EU HTACG adopts General Principles on AI use in JCA dossier preparation — Health Technology Developers remain fully accountable for the completeness, validity, and methodological rigour of any AI-assisted evidence submitted. PDF ↓

№15 15 Jul 2026 EU consultation on classifying high-risk AI under the AI Act closes 23 July — draft guidelines read the conformity test broadly, pulling more medical devices and health pricing tools into scope. The Omnibus deferred the deadlines, not the scope. PDF ↓

№14 8 Jul 2026 Rentosertib enters Phase III — the first drug where AI identified the target, designed the molecule, and guided clinical development end-to-end. HTA frameworks for AI-originated drugs are now urgently needed. PDF ↓

№13 1 Jul 2026 EU Council formally adopts AI Act Omnibus — high-risk AI deadlines confirmed, medical devices extended to August 2028. Transparency obligations remain 2 August 2026. PDF ↓

№12 24 Jun 2026 MHRA National Commission finalising AI medical device regulatory framework and NICE formalising AI evidence-generation guidance — the UK's AI health governance standards are being written now. PDF ↓

№11 17 Jun 2026 Anthropic's Claude Fable 5 launched and pulled by US government directive within three days — the first government-mandated takedown of a publicly deployed frontier AI model, raising new vendor risk questions for regulated AI workflows. PDF ↓

№10 9 Jun 2026 More than 200 AI-discovered drugs now in clinical development — 94 in Phase 1, 56 in Phase 2, 15 in Phase 3 — with analysts projecting a 60% probability that the first AI-discovered drug receives FDA approval in 2027 or 2028. PDF ↓

№9 3 Jun 2026 Nature publishes Robin — a multi-agent AI system that autonomously identified a novel drug repurposing candidate for macular degeneration, analysing 551 papers in 30 minutes with a 200-fold reduction in scientist time. PDF ↓

№8 26 May 2026 NIHR i4i £10m funding call closes — the direct-funded route for companies with AI-enabled diagnostics holding a NICE Early Value Assessment to generate the RWE required for full NICE guidance and NHS-wide reimbursement. PDF ↓

№7 20 May 2026 European Commission's stakeholder consultation on AI transparency obligations closes 3 June — disclosure for interactive AI, deepfake labelling and biometric categorisation under Article 50 of the EU AI Act. PDF ↓

№6 12 May 2026 EU AI Act high-risk deadline confirmed — 2 Dec 2027 for standalone systems, 2 Aug 2028 for medical devices. Pharma and healthcare AI finally have a definitive planning horizon. PDF ↓

№5 6 May 2026 EU AI Act's high-risk obligations are scheduled for 2 Aug 2026 — 13 weeks away. A Digital Omnibus proposal may push this to Dec 2027, but it isn't formally adopted: 2 August remains the current legal obligation. PDF ↓

№4 28 Apr 2026 New UK clinical trial regulations came into force today — post-Brexit regulatory divergence from EU frameworks is now a live consideration for AI-enabled diagnostics seeking UK market entry. PDF ↓

№3 21 Apr 2026 Novo Nordisk partners with OpenAI to accelerate drug discovery — as drugmakers increasingly turn to AI partnerships at scale, the implications for evidence generation and HTA submissions are growing harder to ignore. PDF ↓

№2 14 Apr 2026 Eli Lilly commits $2.75bn to Insilico Medicine for AI-discovered candidates — the first compound from target identification to Phase I in under 30 months, raising questions for HTA bodies evaluating shorter evidence histories. PDF ↓

№1 31 Mar 2026 Pilot issue — Insilico's ISM001-055 becomes the first AI-designed drug against an AI-discovered target to show positive Phase IIa results; over 60% time reduction from initiation to preclinical candidate. PDF ↓

Current AI news meets peer-reviewed research — every week.

Earlier issues