On 7 May 2026, CeSIA took centre stage at Geneva Cyber Week, co-organised by the United Nations Institute for Disarmament Research (UNIDIR) and the Swiss government, to address a major global challenge: "Public-interest research in AI safety and cybersecurity: investing today to make tomorrow safer".
Represented by Charbel Segerie, Pauline Charazac, and Markov Grey, CeSIA joined Pablo Rice from the Paris Peace Forum and Kayle Giroud from the CyberPeace Institute for a wide-ranging roundtable discussion.
We highlighted several troubling trends, including a reported 72% rise in AI-assisted cyberattacks in 2025 and the leak of more than 300,000 ChatGPT credentials on the dark web over the past year — all of which underscores the urgency of action. Against this backdrop, the discussion centred on two key questions: how can the cybersecurity and AI safety research communities be brought closer together? And what initiatives could accelerate the production of policy-relevant research?
The exchanges highlighted the lack of robust evidence and chronic underfunding that hamper effective responses to AI-related cyber threats. Beyond their economic costs, these threats erode public trust and jeopardise international stability. Participants shared ideas for mobilising civil society, increasing research agility, and leveraging mechanisms such as the Common Good Cyber Fund.
With the French G7 presidency and the Geneva AI Summit on the horizon, participants called for concrete steps to make AI safety and cybersecurity genuine priorities:
- CeSIA stressed that Mythos was just one warning signal among a convergence of indicators pointing to growing risks. Holding that AI risks are inherently cross-border in nature, CeSIA underlined the need for proactive government action at the global level. We emphasised the importance of targeted investment in cyber research, in particular through the European Frontier AI Forum.
- The CyberPeace Institute reaffirmed its commitment to supporting NGOs in their cybersecurity research to counter AI-powered attacks.
- The Paris Peace Forum called for action by contributing to the ongoing consultation on AI safety and cybersecurity, conducted in partnership with CeSIA and comprising two separate questionnaires:
- One aimed at public-interest researchers working on AI safety/security and cybersecurity: https://ppfcesiaconsultation1.netlify.app
- One aimed at institutional and philanthropic funders active in this area: https://ppfcesiaconsultation2.netlify.app