Tencent AI chief warns biology risk nears threshold as Hy4 weights prep
Yao Shunyu said distinguishing offensive from defensive biological capability may be harder than in cybersecurity, according to SCMP.
Tencent chief AI scientist Yao Shunyu warned on November 8, 2026, that artificial intelligence's biological capabilities are approaching a dangerous threshold, and said the company will soon publish the model weights and a safety system card for its flagship model Hy4, according to the South China Morning Post.
Speaking at a symposium co-hosted by the Hong Kong Academy of Sciences and the Future Science Prize, Yao said AI could speed up the development of drugs and treatments, but could also give malicious users a means to develop biological weapons or trigger new pandemics.
He added that separating offensive from defensive capability in biology may be harder than in cybersecurity, where the line between the two is more clearly drawn.
Tencent said the training of Hy4 excluded offensive hacking material and instead used defensive data such as vulnerability analysis and patching. Its own safety evaluation found no serious concerns over the model's cyberattack capability, and found that Hy4's ability to turn biological knowledge into real harmful experimental procedures fell below high-risk thresholds, the company said.
Alongside the model weights, Tencent said it will publish a system card containing its cyber and biological safety evaluations. The company released Hy4, its flagship AI model, on November 4, 2026.
Quick answers
What did Tencent's chief AI scientist warn about?
Yao Shunyu said AI's biological capabilities are approaching a dangerous threshold and could give malicious users a means to develop biological weapons or trigger new pandemics.
What is Tencent publishing for Hy4?
Tencent said it will publish Hy4's model weights and a system card containing cyber and biological safety evaluations.
How did Tencent evaluate Hy4's risks?
Tencent said it found no serious concerns over Hy4's cyberattack capability and that its ability to turn biological knowledge into harmful experimental procedures fell below high-risk thresholds.