Artificial intelligence researchers and entrepreneurs gathered in Hong Kong this July 2026 to argue that scientific collaboration can outlast geopolitical friction — even as Washington and Beijing escalate export controls and investment restrictions that have turned AI into a flashpoint of strategic rivalry.

A counter-current to geopolitics

The event brought together a diverse cohort of AI innovators and entrepreneurs to examine how cross-border knowledge-sharing could build an entrepreneurial ecosystem that transcends deepening national divides. Participants framed civilian-to-civilian technological exchange as a durable force that government-level disputes cannot easily extinguish.

Pak-Sun Ting, co-founder and CEO of Votee AI — developer of the first open-source Cantonese large language model (LLM) — described scientific progress as a 'small undercurrent to help go against geopolitics.' 'Geopolitical winds change from time to time,' Ting said. 'Government to government, there could be some disagreements. But from civilian to civilian, technology to technology, as long as they allow us and it is doable, we will take it and work with it.'

The policy backdrop

For policymakers in Beijing and Washington, AI has become a high-stakes arena tightly bound to national security and strategic interests. US export controls on advanced semiconductors and heightened scrutiny of foreign investment have framed the technology as a zero-sum contest — where one nation's capability gain is perceived as a direct threat to the other.

That binary framing stands in sharp contrast to the view held by many working scientists and developers, who argue that the field's greatest advances have historically emerged from open, international exchange. China President Xi Jinping has also publicly called for 'openness' in AI development, even as US officials accuse Beijing of leveraging open collaboration asymmetrically.

Why it matters

The gathering drew participants connected to institutions and organisations spanning Silicon Valley, Stanford University, Toronto, Kuala Lumpur, and beyond — including figures associated with freeCodeCamp.org, represented by founder Quincy Larson, and technologist John Whaley. The breadth of participation underscored that practitioner-level networks remain active even as official channels narrow.

Open-source language models — such as the Cantonese LLM built by Votee AI — represent one concrete domain where cross-border collaboration produces tools that serve linguistically underrepresented communities globally, a use case that sits awkwardly within purely nationalist frameworks for AI governance.

What's next

The durability of practitioner collaboration will depend heavily on how far export-control regimes expand and whether visa and data-transfer restrictions tighten further across major AI hubs. Companies like Anthropic, headquartered in Silicon Valley, operate in an environment where even routine research partnerships now require legal review for compliance with trade rules.

As the US-China tech rivalry intensifies, the informal networks forged at gatherings like this Hong Kong summit may become increasingly important pressure valves — and increasingly scrutinised by regulators on both sides.