A presentation by Murielle Popa-Fabre, PhD (École Normale Supérieure, Council of Europe). This event is part of the Digital Governance Research Colloquium hosted by the Centre for Digital Governance.
Recent advances in LLM-powered applications are expanding to multimodal data aggregation, yielding an unprecedented form of behavioural predictability. Systems embedding LLMs better capture human behaviour through advanced simulation techniques, with implications for data protection, private life, dignity and autonomy. Analysing these developments is essential for creating future-proof governance measures addressing emerging privacy and human rights risks.
After analysing six key technological developments that fundamentally alter the privacy and human rights landscape, we discuss their implications in an era of enhanced behavioural predictability. We can observe the early signs of a gradual shift from an attention economy - where human attention is traded through advertising - to an intention economy, where technological actors use multimodal signals to forecast human intent through highly adaptable 360° individual user profiles, creating both a seamless technological interaction and a "predictive and adaptable data-caging". While current system adaptiveness leaves users feeling understood and assisted this new configuration is fundamentally questioning the human right to autonomy and private life.
Integrating these capabilities with smartphone multimodal data (sensors, geolocation, usage patterns, etc…) opens unprecedented behavioural prediction possibilities that are not only challenging traditional privacy assumptions but open the way to infer users' mental states. Recent research already demonstrates the possibility of integrating brain data into LLM prompts to partially decode brain activity (so-called Brain LLMs). Understanding these developments is essential to grasp how these systems can shape human actions and impact human autonomy.
In summary, the technological convergence of (1) new multimodal data aggregation, (2) efficient predictive and pattern-finding abilities, (3) fundamental adaptability, and (4) feeding brain waves data into LLMs, positions 2025 as a pivotal moment for Human Rights and privacy risk assessment.
Keywords: Large Language Models (LLMs), Privacy Risk Assessment, Behavioural Prediction and Simulation, Multimodal Models, Human autonomy, smartphone multimodal data, Neurotech wearables, Brain LLMs.
Murielle Popa-Fabre (PhD): After an international academic career in Natural Language Processing and Cognitive Neuroscience, teaching in France and at Cornell (NY), Murielle Popa-Fabre (PhD) currently advises on AI impact on Human Rights, data privacy and the Judiciary at the Council of Europe, and on Generative AI responsible development in UN agencies (ITC-ILO) and the private sector (media, legal, banking).
As a consultant, she advises C-level decision makers on Generative AI corporate strategy and the National innovation investment fund France 2023 on Generative AI ecosystem building. She actively disseminates technical understanding of Large Language Models, AI regulation and neuro-technologies in media and policy contexts.
She is Senior Tech and Policy Advisor at the AI & Society Institute at École Normale Supérieure (ENS-PSL Paris), and ML & NLP Expert for the Council of Europe on Human Rights, Privacy and Freedom of Expression (CAI Huderia, CDMSI, TP-D), contributing to AI Governance training (Huderia Academy) and co-rapporteur of the Guidelines on Generative AI and Freedom of Expression.
Recently authored Tech & Policy reports:
- Expert report “Data Protection in the Context of Large Language Models” presented by Isabel Barberá and Murielle Popa Fabre, June 2025
- Action research report “Action research on the impact of neurotechnology on lifelong learning – and how to harness it for Social Justice” ITC ILO, September 2025
Registration is required for this event.
Photos and videos will be taken during this event to capture the atmosphere and document our event. By participating, you also agree that we may use your images and videos for marketing and promotional purposes, including publication on our website, on our social media, in printed marketing materials and in press releases. Our partners involved in the financing of this event and speakers may also have access to this content. If you agree to these terms and conditions and wish to participate in our event, you do not need to do anything else. Your participation constitutes consent to the use of your images and videos as described above.
Please note that your participation in the event is not conditional on your consent to the use of your images and videos. However, if you do not want your images and videos to be used, please let us know and speak to us directly at the event and we will take appropriate measures to protect your privacy. You can find our privacy policy here. Thank you for your understanding and support.