Tencent's chief AI scientist Yao Shunyu warns AI's cyber and soon bio abilities will reach a threshold for harm
Speaking at a science symposium in Hong Kong on Saturday, the former OpenAI researcher said open models could speed drug discovery for small labs but also help people build biological weapons.
Why it matters
Warnings about AI-enabled biological risk usually come from US labs such as Anthropic and OpenAI, which have tightened safeguards around their bio-capable models. Yao Shunyu, a former OpenAI researcher who now leads AI research at Tencent, told a science symposium in Hong Kong on Saturday that 'the cyber abilities of models and soon the bio ability of models will reach a threshold where it's actually possible to do harm.' His remarks matter because Tencent and other Chinese firms release open-weight models that anyone can download. Watch whether Chinese labs adopt bio-misuse evaluations or release restrictions.
Key facts
- Yao Shunyu is Tencent's chief AI scientist and ex-OpenAI
- Spoke at a science symposium in Hong Kong on Saturday
- Says model cyber abilities are near a harm threshold
- Says bio abilities will soon reach it too
- Open models could speed drug discovery for small labs
- He warned they could also help build bioweapons