As Europe aims for "digital sovereignty", biomedical agentic AI could be the next big field

Key Concepts

• Digital Sovereignty: Europe's aim to reduce reliance on US and Chinese tech giants and develop its own digital solutions.
• AI Race: The global competition between major powers (US, China, Europe) in the development and application of Artificial Intelligence.
• Agentic AI Infrastructure: A system designed to integrate, harmonize, and reason over biological data, aiming for "biological super intelligence."
• Biological Super Intelligence: An advanced form of AI capable of understanding complex biological processes, disease causes, and predicting outcomes, surpassing human cognitive abilities.
• LLMs (Large Language Models): AI models trained on vast amounts of text data, capable of understanding and generating human-like text. Examples include GPT-4 and OKIM's "Zero."
• Federated Learning: A machine learning technique that allows models to be trained on decentralized data sources without the data leaving its original location, preserving privacy.
• Data Privacy vs. Innovation: The ongoing debate in Europe regarding the balance between protecting individual data and enabling large-scale data aggregation for scientific advancement.

European Digital Sovereignty and the AI Race

The French president, Emmanuel Macron, recently attended a summit in Berlin focused on digital sovereignty, highlighting Europe's strategic imperative to catch up with the US and China in the global Artificial Intelligence (AI) race. Macron, alongside German counterpart Friedrich Mertz and other EU leaders, called for increased European innovation to counter the dominance of American big tech and Chinese companies. This initiative saw pledges of over 12 billion euros in investments for the digital sector. The summit precedes a proposed EU rollback of regulations on AI and data protection, signaling a shift towards fostering innovation.

Macron emphasized Europe's desire to "design our own solutions" rather than being a "client of the big entrepreneurs or the big solutions being provided either from US or from China." He articulated a "two-fold challenge": to innovate for competitiveness while simultaneously protecting European data, infrastructure, and sovereignty.

OKIM's "Agentic AI Infrastructure" for Biological Super Intelligence

A key participant at the summit was OKIM, a biotech company that announced a new project to build a pan-European infrastructure for biological data, making it "AI ready." Tom Clausel, CEO of OKIM, explained the concept of "biological super intelligence" as a necessary evolution to tackle the immense complexity of biological systems, which currently eludes human understanding, particularly in understanding diseases like cancer and Alzheimer's.

Project Details and Goals:

• Collaboration: OKIM is partnering with two leading French and German cancer research centers: Gustave Roussy in Paris and the Charité Comprehensive Cancer Center in Berlin.
• Infrastructure: The project aims to establish an "agentic interface" for data integration and harmonization.
• Reasoning Layers: This interface will incorporate new reasoning capabilities and Large Language Models (LLMs) to create an "artificial super agent."
• Primary Objective: To automate and improve decision-making in research and patient care.
• Research Focus: Enhancing clinical trials, generating new hypotheses, and testing novel ideas.
• Long-Term Goal: To achieve improved development and patient-side decisions, ultimately leading to a "causal intelligence" or "super intelligence" that surpasses current human capabilities in understanding disease causes and predicting responses.

Clausel highlighted the changing demographics of diseases, such as younger patients with pancreatic cancer, as evidence of the need for new analytical approaches, as current understanding is insufficient. The initial step involves consolidating and harmonizing data, followed by applying advanced LLMs and reasoning models.

The Role of AI in Drug Development and Healthcare

The discussion addressed the current state of AI in biology and healthcare. While AI has contributed to rapid developments like COVID-19 vaccines, Clausel argued that it hasn't yet achieved the breakthrough of developing entirely new drugs or understanding complex diseases like Alzheimer's or the root causes of cancer. He cited the development of drugs like Ozempic as significant for productivity but noted that the underlying technology wasn't primarily AI-driven.

Clausel believes that while AI is present in many drug development pipelines, it has not yet delivered on the promise of treating diseases where traditional pharmaceutical approaches have failed. He attributes this to a lack of sufficient data and the absence of the right reasoning models and LLMs to achieve "super intelligence."

Europe's Position in Biomedical AI and LLMs

Regarding Europe's standing in biomedical AI compared to the US, Clausel acknowledged the impressive capabilities of current LLMs like GPT-4, GPT-5, and Gemini 2.5, which are trained on global web data. However, he pointed out their limitation in leveraging curated patient data from specialized centers.

OKIM's strategy involves developing "Zero," its own LLM, which has demonstrated superior performance on certain benchmarks compared to GPT-4 in recent research. Clausel emphasized the importance of training LLMs on the "right data" and incorporating "reinforcement learning" from new lab data, platform usage, and patient data to drive innovation.

He also noted that major players like OpenAI (with its company Virtual Bison), Entropic, and Oracle are actively investing in AI and biology, suggesting that future pharmaceutical giants might be "AGI driven LLM driven." OKIM believes that agentic processes, new LLMs, reasoning models, and superior data will enable them to outperform traditional big pharma with fewer resources.

Digital Sovereignty and Data Privacy in Biological Data

The sensitivity of biological data, directly related to patient health, makes Europe's pursuit of digital sovereignty in this field particularly crucial. Clausel acknowledged that while the US is advancing rapidly in medical biology, no single entity has yet achieved dominance. He sees an opportunity for Europe to secure a "winning place" in this domain, especially concerning data.

Data as the Key:

• Data Requirements: Access to large volumes of curated patient data and laboratory data is essential for AI development.
• Privacy Concerns: Data privacy is a significant consideration. OKIM has developed federated learning technology to enable remote data access while preserving privacy.
• Balancing Privacy and Innovation: Clausel argued that the discourse on data privacy in Europe might be overly focused on individual protection, potentially hindering innovation. He suggested that aggregating European patient data could lead to significant discoveries.
• Potential Risks of Over-Regulation: While acknowledging the existence of risks associated with data breaches, Clausel posited that excessive emphasis on individual data privacy could be "dangerous" by slowing down innovation, especially in the face of intense global competition. He raised the question of whether Europe should reconsider its approach to data gathering at scale to accelerate progress against the US and China.

The discussion concluded with the acknowledgment that while individual data privacy is important, the potential benefits of large-scale data aggregation for medical advancement warrant careful consideration and potentially a reevaluation of current political approaches.