International policymakers are accelerating efforts to establish binding governance frameworks for artificial intelligence, with open-source models emerging as a central point of contention in negotiations now underway in Geneva. The talks, convened under the auspices of the International Telecommunication Union, have drawn delegations from more than 60 countries and mark the most ambitious attempt yet to create a unified approach to AI oversight.
At the heart of the debate is whether freely available AI models — which can be downloaded, modified, and deployed by anyone with sufficient computing resources — should be subject to the same safety assessments and transparency requirements as their proprietary counterparts. Advocates for open-source AI argue that broad accessibility accelerates innovation and democratises the technology, while critics warn that unrestricted distribution could enable misuse at scale.
“We are navigating entirely uncharted territory,” said Dr. Elena Vasquez, director of digital policy at the Centre for Responsible Technology, a Brussels-based think tank. “The regulatory instruments we developed for social media and data privacy simply do not map onto the unique characteristics of generative AI systems — particularly those released under open licences.”
Industry figures have signalled a willingness to cooperate while pushing back against what some describe as overreach. A coalition of 14 technology companies, including several prominent European AI labs, issued a joint statement last week calling for “risk-proportionate regulation” that distinguishes between frontier models and smaller-scale applications. The statement also urged negotiators to avoid measures that would effectively mandate centralised control over model distribution.
The discussions are being closely watched by investors, who have poured an estimated $27 billion into AI startups during the first half of 2026 alone. Market analysts at Goldman Sachs noted in a client brief this month that clarity on the regulatory landscape could unlock significant additional institutional capital currently sitting on the sidelines. “Uncertainty is the enemy of investment,” the brief stated. “Once the rules of the road are established, we expect a substantial reallocation toward AI infrastructure and application-layer companies.”
Negotiators are working against an informal deadline of September 2026, when a draft text is expected to be presented to member states. Whether the resulting framework will prove sufficient to satisfy both safety advocates and innovation proponents remains an open question — one that is likely to shape the trajectory of AI development for the remainder of the decade.