FreeLLMTencent to launch open-weight model Hy4 amid AI bioweapon warnings
Tencent plans to release Hy4 as an open-weight model with safety report, while recent Hugging Face breach highlights open models' role in security.
Tencent's chief AI scientist warned that AI could soon be used to create bioweapons and said that neither fully open nor fully closed models alone can solve the risk.
The company announced that its upcoming model, Hy4, will be released with open weights and an accompanying safety report, claiming its biological capabilities are below a high‑risk threshold.
Hy4's training data were filtered to remove hacking capabilities, a measure the article says survives attempts to strip safety filters (abliteration).
In July 2026, Hugging Face disclosed an intrusion driven by an autonomous AI agent system; the forensic analysis was performed on the open‑weight model GLM‑5.2 because hosted models blocked the necessary data.
The incident showed that open models can aid defenders when safety guardrails on commercial APIs prevent analysis, but also underscored the difficulty of controlling model misuse.
Ce que cela signifie pour vous
For users, the release of Hy4 means a new free, open-weight option that may be suitable for local deployment, provided its safety limits meet their needs. The Hugging Face breach illustrates why having a capable, self‑hosted model like GLM‑5.2 is valuable for security teams, while also reminding the community that open models can be repurposed for offensive tasks if not properly guarded.
Sources
- Tencent's AI chief says AI bioweapons are coming, but closing off models isn't the answer, The Decoder, 9 Nov 2026
- Security incident disclosure — July 2026, huggingface.co (official), 16 Jul 2026
Rédigé automatiquement à partir des sources ci‑dessus, puis vérifié par notre code : chaque nombre et chaque nom doivent y figurer, et aucune phrase ne doit copier une source. Une erreur ? Signalez‑la.
Quelle IA pouvez‑vous utiliser gratuitement aujourd’hui ?
Tous les modèles IA gratuits, classés par votes humains anonymes, avec la mémoire et la vitesse dont ils ont besoin sur votre propre GPU.