Early-warning capacity
Signals, methods and workflows to detect information voids, volume anomalies and coordinated manipulation before they become systemic crises.
Cognitive Resilience Institute
Cognitive resilience for democratic societies in the age of frontier AI.
Building public-interest capacity to strengthen the integrity of democratic information environments as synthetic media, general-purpose models and AI agents reshape public trust.
Read the mission
Why now
The information environment is becoming synthetic, adaptive and agentic.
Democratic societies rely on shared signals: public debate, journalism, civic participation, institutional trust and the ability to tell authentic coordination from artificial amplification.
Frontier AI changes the cost, scale and realism of manipulation. Synthetic media can imitate people. General-purpose models can generate persuasion at scale. AI agents can coordinate, test and adapt narratives faster than human institutions can respond.
The challenge is not only misinformation. It is whether societies can remain cognitively resilient when the signals they depend on become easier to simulate.
Frontier AI safety
Frontier AI safety does not end at the model boundary.
Model evaluations matter. But democratic risk emerges when powerful models interact with real information systems: feeds, media ecosystems, political debate, civic participation, elections and public institutions.
A model that is safe in isolation can still contribute to systemic harm when it enables synthetic influence, automated trust-building, coordinated inauthentic behaviour or manipulation of public attention.
CORE focuses on this social layer of AI safety: how frontier systems affect the integrity of the information environment that democratic societies depend on.
Mission
To make democratic societies harder to manipulate.
CORE is an independent public-interest institute for cognitive resilience and information integrity. We develop research, tools and standards to help societies detect early signs of coordinated manipulation, understand emerging AI-enabled risks and strengthen the civic capacity to respond — without undermining openness, pluralism or democratic debate.
Resilient societies do not need to control information. They need the capacity to understand, verify and withstand manipulation.
What we are building
Public-interest capabilities, built to be shared.
AI-enabled influence risk
Evaluation of how general-purpose models, synthetic media and agentic systems can affect public discourse, trust and democratic information environments.
Open public-interest tooling
Shared technical capabilities that researchers and institutions can audit, adapt and extend without creating new dependencies.
Cognitive resilience methods
Research and educational approaches that help societies recognise manipulation, preserve pluralism and build durable civic immunity.
How it works
From early signals to civic strength.
Sense
Monitor public information environments for early signals of manipulation, information voids and synthetic amplification.
Evaluate
Assess how AI systems, synthetic media and coordinated behaviour may create systemic democratic risk.
Explain
Translate complex technical and social signals into clear, decision-useful intelligence.
Equip
Build open methods, tools and standards that institutions and civil society can use.
Strengthen
Help democratic societies become more resilient without compromising openness or fundamental rights.
Principles
Built for public trust.
Independence
Designed to serve democratic resilience, not political, commercial or platform interests.
Openness
Transparent methods and open capabilities wherever openness strengthens accountability and public value.
Scientific and technical rigour
Research discipline, technical evaluation and information-integrity methodology.
Democratic pluralism
Resilience must protect open debate, not narrow it — integrity of the environment in which speech circulates, never control of speech.
European public interest
A contribution to technological sovereignty, institutional readiness and democratic resilience.
Human agency
The future of information integrity depends on citizens, journalists, researchers and institutions retaining the ability to interpret and act.
Who it serves
For the people and institutions that keep democracy legible.
Journalists and researchers
Clear signals, methods and evidence to investigate emerging manipulation and AI-enabled influence.
Civil society
Tools and knowledge to build resilience before crises become visible.
Democratic institutions
Independent capacity to understand systemic information risks and respond proportionately.
Citizens
A healthier information environment where public trust is harder to simulate and easier to earn.
A different kind of AI safety
The civic layer of frontier AI safety.
The next generation of AI risk will not appear only as model failures, benchmark gaps or isolated misuse. It will appear in the social systems where information becomes belief, belief becomes coordination and coordination becomes democratic power.
CORE exists to build that civic layer: independent, open, rigorous and oriented toward societies that remain free because they remain resilient.