Tencent AI chief warns models will soon pose biology risks, SCMP reports
What happened
Tencent's chief AI scientist, Yao Shunyu, warned at a science event in Hong Kong that AI models will soon be able to cause harm in biology, similar to how they can already be used in cybersecurity. He highlighted the need for strict safety measures, particularly for open-source AI models. Yao made these comments just days after Tencent launched its new Hy4 AI model on November 4. Tencent stated that Hy4 has strong biology knowledge but limited ability to design experiments, and the company plans to release the model's weights and a safety report soon.
Key details
- Yao Shunyu is Tencent's chief AI scientist and previously worked at OpenAI.
- The Hy4 model was launched by Tencent on November 4.
- Hy4 was trained without material on exploiting software flaws and focused on defensive security data.
- Hy4 is rated as the second-best Chinese AI model by benchmarking firm Artificial Analysis.
- Tencent plans to release Hy4's model weights and a safety testing report in the coming days.
Why it matters
Chinese technology companies developing open AI models may need to implement stricter safety protocols to address emerging biological risks. Organizations and developers using or releasing open-source AI models could face new requirements or limitations to prevent misuse in biological contexts. The planned release of Hy4's model weights and safety report will provide transparency for researchers and regulators assessing the model's potential risks and safeguards.