Issue 7
Novo Nordisk to use Anthropic's Claude in drug research
Danish company Novo Nordisk and Anthropic announced a collaboration on 16 September 2026 to speed up drug discovery. Novo said the AI tools open up "entirely new scientific possibilities".
Who is affected and how
- pharmaceuticals
- A major drugmaker is publicly building its research on a large language model, which gives competitors a benchmark for their own AI strategies.
Von der Leyen: healthcare AI should be European and must not replace doctors
On 16 September 2026, Ursula von der Leyen put AI-powered healthcare on the EU agenda and called for a "European" solution. According to reports, the Commission plans initiatives to support AI…
health policyAhead of the Commission's November initiatives, it is worth mapping the national position and one's own healthcare AI programmes.health ITThe expectation of a "European" solution may affect vendor and data-storage decisions, so procurement plans should be aligned with it.doctorThe emphasis is that AI will not replace doctors, so for professional chambers, securing the supporting role of these tools is the negotiating position.regulation · United KingdomUK commission publishes 44 recommendations on regulating AI in healthcare, from pre-market to lifecycle
The UK's National Commission published 44 recommendations on regulating AI in healthcare on 14 September 2026, proposing a staged framework. The recommendations shift the emphasis from pre-market…
health policyThe staged regulatory model can serve as a template for other countries' frameworks.medtech companyLifecycle-based oversight may foreshadow post-market monitoring obligations.health ITDeployers are likely to face a requirement for continuous performance monitoring.research · globalEvidence for medical AI effectiveness is lacking: only 34 of 1,357 FDA-authorized devices registered for prospective studies
A PLOS Digital Health analysis published in August reviewed 1,357 AI/ML devices authorized by the FDA up to 5 December 2025 and found that most had not been clinically tested; only 34 were registered…
doctorRegulatory authorization alone is not evidence of clinical effectiveness, so when choosing a device it is worth asking for prospective data.hospital managerProcurement should require real-world clinical validation.medtech companyThe lack of prospective evidence is a market and credibility risk.health policyPressure for regulatory requirements on proof of effectiveness may grow.regulation · United StatesUS health officials deploy medical AI as safety concerns persist
The FDA has so far authorized 1,524 AI devices, but no generative AI product for mental health (14 September 2026). According to Haider Warraich, a pilot run by the Trump administration's health…
health policyThe tension between state-level deployment pilots and federal oversight is a regulatory risk.doctorThe move toward autonomous AI raises questions of liability and oversight.medtech companyNo generative AI product for mental health has yet been authorized, which is both a market gap and a regulatory risk.Hungary
More key stories of the week
Oracle Health's clinical AI agent aims to reduce nurses' documentation burden
Oracle Health unveiled its clinical AI agent on 14 September 2026, designed to ease nurses' documentation burden and streamline care. The company is thereby targeting the nursing workflow…
hospital managerInstitutions working in the Oracle environment should assess the rollout and cost terms of the new tool.health ITIntegration of the agent and EHR data flows need to be prepared.drug development · globalGenScript and Twist join Eli Lilly's TuneLab AI platform
On 16–17 September 2026, GenScript Biotech and Twist Bioscience both joined Lilly TuneLab, a collaborative AI-based drug discovery platform. GenScript is contributing large-scale wet-lab validation…
pharmaceuticalsUnder the TuneLab model, contract partners supplying experimental data become integral parts of AI platforms, which matters when selecting CRO partners.drug development · AsiaByteDance AI drug discovery unit Anew Labs raises $290M at $1.5B valuation
Anew Labs, the AI drug discovery unit ByteDance spun off, closed a $290 million funding round on 16 September 2026 at a valuation of $1.5 billion. The funding came after the company's spin-off.
pharmaceuticalsA technology giant's spun-off AI unit is entering drug discovery with substantial capital, a matter for partnership planning and competitor monitoring.research · globalVirtual biotech puts up to 37,000 AI agents to work on drug discovery, with lung cancer candidate
A "virtual biotech" described in Science and Nature uses thousands of AI scientist agents for drug discovery and development (17 September 2026). According to Nature, the system also identified a…
pharmaceuticalsThe multi-agent framework points to automation of target identification and development planning, which makes pilot testing of research workflows worthwhile.researchStudies find AI scribes cut doctors' documentation time and burnout
University of Toronto research and further analyses find that ambient AI scribes reduce doctors' documentation time and may ease burnout (15–17 September 2026). One report says they save 34 minutes a…
doctorThe time savings can be backed with data, so they provide a reference point in applications for institutional rollout.hospital managerThe return on investment and the effect on burnout can be argued in the purchasing decision.health ITIt is worth defining the evaluation metrics in advance.research · globalDrug firms' proprietary data boost AI protein models
Six sources report that drug companies' non-public data significantly improves AI-based protein models (14 September 2026).
pharmaceuticalsProprietary data is a competitive advantage in model training, so data-sharing and federated learning collaborations are worth considering.tool · European UnionTandem Health raises $100M Series B, led by EU's €5bn tech startup fund, for European expansion
Tandem Health announced a $100 million Series B round on 14 September 2026, led by the EU's €5bn tech startup fund. It will use the money to expand across Europe and to build an "AI-native clinical…
health ITA European ambient AI provider is expanding with substantial capital, so it may become a new supplier alternative to US solutions.doctorThe medical assistant may extend to a wider range of the clinical workflow.research · United StatesNearly one in five young people turns to AI for mental health advice
Surveys indicate that almost one in five young people turn to AI for mental health advice (15–17 September 2026). One report says many teenagers treat AI as a therapist.
doctorWhen questioning young patients, it is worth mapping their AI use as well.health policyWidespread chatbot use among young people strengthens the case for regulation.research · AsiaAlibaba open-source AI model detects nearly 150 conditions, outperforming 23 of 26 radiologists
Alibaba has released as open source an AI model that detects nearly 150 conditions from CT images (18–19 September 2026). In results published in Science, 23 of 26 radiologists performed worse than…
doctorOpen models that outperform radiologists warrant local validation.medtech companyAn open-source model with broad coverage may put pricing pressure on closed vendors.regulation · United StatesFDA order keeps radiology AI under 510(k) premarket review
According to two trade sources, an FDA order keeps radiology AI under 510(k) review (17–18 September 2026). A separate analysis says the regulatory risks are already present.
medtech companyRevenue for radiology AI remains tied to premarket clearance, so the 510(k) plan and documentation need to be maintained.doctorCheck the authorization status of radiology AI tools before adopting them.regulation · United StatesConnecticut bans AI-based health insurance denials for state employees
Connecticut announced AI rules for state employee health plans between 16 and 18 September 2026. The rule bans AI-based health insurance denials.
health policyA state-level AI rule could serve as a model for other states.doctorDoctors can cite the rule against denials that rely on AI.clinical deployment · United States25 health systems build AI agents with Epic; Epic announces real-time prior authorization
Twenty-five health systems are building AI agents in Epic's "Agent Factory" program (15 September 2026). According to Suki's chief executive, Epic's real-time prior authorization announcement also…
health ITAn EHR vendor's own agent platform also affects the position of independent suppliers.hospital managerIn an Epic environment, the role of third-party scribes needs rethinking alongside in-house agent development.