.png){kind=logo}
Tencent’s chief AI scientist, Vinces Yao Shunyu, has warned that artificial intelligence is approaching a point where it can cause serious harm in biology. He made the remarks at a science symposium in Hong Kong, co-hosted by the Hong Kong Academy of Sciences and the Future Science Prize.
Yao said, “Frontier AI systems are advancing rapidly in biological understanding. They could soon help researchers design drugs, identify treatments and solve complex scientific problems at a much faster pace.”
Yao framed the issue as a dual-use challenge. AI can accelerate discovery in medicine and life sciences, but it can also make dangerous knowledge and experimental pathways more accessible.
He compared biology with cybersecurity. In cyber systems, developers can more clearly separate offensive and defensive functions. In biology, that distinction is much harder to draw. At the same time, he cautioned that the same capabilities could assist malicious actors. AI could lower barriers to developing biological weapons or triggering a new pandemic, he warned.
A model that helps scientists understand proteins or develop therapies may also provide useful support for harmful biological work. This overlap makes safeguards more difficult to design and enforce.
Yao paid particular attention to open-weight AI models. These systems allow developers and users to download, modify and run model weights outside a company’s controlled environment.
He said Chinese companies and open-source developers must handle safety and security responsibly. The warning comes as Tencent prepares to release the weights and a safety system card for its flagship model, Hy4.
Yao argued that developers should evaluate a model’s biological capabilities before release. They should also build safeguards that limit harmful assistance while preserving legitimate research uses. His message places biosecurity alongside cybersecurity as a core test for frontier AI. Companies, regulators and research institutions will need to cooperate closely to manage the technology’s benefits without enabling catastrophic misuse.
Also Read: AI-Powered Vulnerability Detection: What GPT-6 Cyber Means for Developers