PolyOMIC Solutions
AI Moves Into Clinical Reality — and Security Becomes the New Battleground
This week in digital health
-
AI + biopharma vertical integration — the shift is no longer about AI tools sold to pharma; but about AI platforms acting as end-to-end engines to generate drug candidates through proprietary foundation models trained on exclusive datasets. This creates a compounding data moat — a self-reinforcing cycle where continuously expanding, high-quality data makes the AI system smarter, faster, and increasingly difficult to compete with over time.
Notable 2026 deals:- Eli Lilly – Chai Discovery — Chai’s algorithm for novel biologic therapeutics design across targets
- GSK – Noetik — OCTO-VC “virtual cell” foundation model simulating tumor biology at cell-state level to predict target response
- Pfizer – Boltz — Boltz-2/BoltzGen AI tools for protein design and biomolecular modeling
GEN Engineering News →
- Anthropic launches AI cybersecurity initiative “Project Glasswing” with Big Tech — partners include Nvidia, Apple, Microsoft, and Google. AI identified thousands of real-world vulnerabilities in a coordinated security research program — a signal that AI is becoming core infrastructure for both attack and defense. Full breakdown below. Anthropic →
- AI moves into clinical decision support — health systems like Mount Sinai are actively piloting AI for clinical reasoning, not just documentation. The trajectory: documentation → clinical support → decision influence. Validation, workflow integration, and regulatory clarity remain the key barriers. Healthcare IT News →
- Digital biomarkers are crossing from exploratory to validated — FDA engagement on wearable-derived biomarkers as clinical endpoints is growing. The regulatory pathway that was theoretical two years ago is now actively being defined. For device and pharma companies, the window to establish validated digital endpoints is open now. Nature Communications Medicine →
- Cybersecurity is becoming a clinical safety issue — as AI systems expand the attack surface in healthcare settings, ransomware and data breaches are no longer just IT problems. They delay care, compromise drug dosing systems, and expose patient data. Security is now patient safety. ComplianceHub →
-
CMS ACCESS Model shifts payment from individual visits to outcome-based reimbursement — the ACCESS Model focuses on patients with chronic conditions across four clinical tracks:
- eCKM — Early cardio-kidney-metabolic conditions
- CKM — Cardio-kidney-metabolic conditions
- MSK — Musculoskeletal conditions
- BH — Behavioral health conditions
Why this matters: CMS is accelerating the shift from fee-for-service to outcomes-based, technology-enabled care at scale, with broad ecosystem participation already underway. Healthcare IT News →
- Foundayo pills are now available via Amazon Pharmacy — Eli Lilly’s newly approved weight-loss pill is now being distributed through Amazon Pharmacy, marking a significant shift toward direct-to-consumer drug access. Reuters →
AI + Cybersecurity just scaled overnight — what it means for healthcare
The launch of Anthropic’s Project Glasswing marks a turning point in cybersecurity. Powered by its frontier model Claude Mythos Preview, the system can autonomously identify thousands of software vulnerabilities — at a scale and speed beyond most human experts.
For healthcare, this is not abstract. It directly reshapes risk. Hospitals, digital health platforms, and life sciences systems sit on some of the most targeted infrastructure in the world. If AI can find vulnerabilities faster, it can also be used to exploit them faster — shrinking the window between discovery and attack.
At the same time, Glasswing signals the beginning of AI-native defense, where the same models that surface vulnerabilities are also used to patch and secure them. Major tech and healthcare infrastructure players are already participating in this controlled rollout.
Bottom line: cybersecurity is no longer a reactive IT function — it is becoming an AI-speed race between detection and exploitation. Healthcare is now firmly inside that race.
Sources: Anthropic Project Glasswing · Reuters · ComplianceHub · STAT News
Anthropic’s Project Glasswing, AI cybersecurity project with Big Tech partners
Picus Security’s analysis highlights that Project Glasswing is not just a cybersecurity breakthrough, but shows that AI can now discover vulnerabilities faster than organizations can fix or control them
Source: Picus Security →
Stay decoded.
Join readers getting the clearest signal in digital health — every Wednesday, free.
No spam. Unsubscribe anytime. Powered by Beehiiv.