WORLDTECHTencent AI chief warns biology risks from models are nearing
Tencent's chief AI scientist, Yao Shunyu, has warned that AI models will soon be able to cause real harm in biology, a risk he says already exists in cybersecurity. He made the remarks at a science event in Hong Kong, as reported by the South China Morning Post.
What Yao said in Hong Kong
Speaking at a science event in Hong Kong on Saturday, Yao said that the cyber abilities of AI models, and soon their biological abilities, would reach a threshold at which harm becomes actually possible, according to the South China Morning Post. The outlet is the only source reporting the remarks.
Yao also argued that biology risks may be harder to manage than cyber risks, a point he made during the same appearance.
Timing alongside the Hy4 launch
The remarks came only days after Tencent launched Hy4, which the company trained without offensive hacking data. The timing places Yao's warning alongside a fresh model release from the company.
Safety duties for open models
Yao said Chinese companies that release open models, which anyone can download and modify, need to take safety very seriously. Open models can be altered after release, which is why he singled out those developers.