Tencent releases open-source Hy4 AI model trained with limited defensive security data

Tencent◼ has released Hy4, an open-source AI model, which it says was trained on limited cybersecurity data focused on defensive content. Training content was filtered to mitigate offensive capabilities. The article gives no CVEs, version details or specific misuse findings, so defenders should evaluate the model's behaviour independently before relying on these safeguards.

For security analysts · Written by AI from 1 source · Wrong?