Tencent strips cyber-hacking capabilities from Hy4 model, following Moonshot AI's move with Kimi K3.1 Pro
Tencent Holdings said it stripped its latest Hy4 model of cyber-hacking capabilities while retaining cyber defence capabilities. Moonshot AI made a similar move with Kimi K3.1 Pro, according to the South China Morning Post.
Tencent Holdings said it removed the cyber-hacking capabilities of its latest Hy4 model while keeping its cyber defence capabilities, the South China Morning Post reports. The step follows a similar move by Moonshot AI with its Kimi K3.1 Pro.
What was reported
According to the SCMP, China's leading AI model companies are beginning to curb the cybersecurity capabilities of their most advanced open-weight models. The report ties the shift to a string of high-profile AI hacking incidents involving both US and Chinese companies.
Tencent made its statement earlier this month. The company said Hy4 no longer has cyber-hacking capabilities but retains cyber defence capabilities. The SCMP says Moonshot AI took a similar step with Kimi K3.1 Pro.
The SCMP article, headlined "Can Chinese AI companies curb cyber risks from open-weight models without losing their edge?", frames the central question as whether these restrictions can be applied without costing the models their competitive position.
What this means
Tencent and Moonshot AI are both separating offensive from defensive cybersecurity capability in their open-weight models, rather than treating cybersecurity as a single capability to keep or remove. Tencent's statement says defence stays while hacking is removed. The SCMP's framing shows that the open question is whether this approach preserves the models' competitiveness.
Related Articles
Tencent AI chief warns models will soon pose biology risks, days after Hy4 launch
Tencent chief AI scientist Yao Shunyu said AI models' cyber abilities, and soon their biology abilities, will reach a threshold where harm is possible, according to the South China Morning Post. He spoke days after Tencent launched its flagship Hy4 model on 4 November.
Anthropic Warns Small AI Models Are Months From Self-Replicating Botnets; Experts Call Threat Familiar
Anthropic's Frontier Red Team published research suggesting small AI models may be only months away from replicating across insecure devices as autonomous botnets. Experts told Engadget the capability is plausible, but that the framing overstates how new the threat is.
Analysis finds Blackwell fingerprints in six open-weight models from four Chinese labs
Unemployment Labs reports that six models released since April 2026 by DeepSeek, Moonshot AI, MiniMax and Ant Group carry numerical patterns matching NVIDIA Blackwell GPUs. Earlier models from the same labs match the older Hopper generation. The analysis says it does not establish where the chips were located or whether export controls were violated.
VulnCheck: Project Glasswing has 37,966 candidate vulnerabilities, 217 marked fixed in public records
Security firm VulnCheck reports that seven months after launch, Anthropic's Project Glasswing has produced 37,966 candidate vulnerabilities, with 217 marked fixed in public disclosure records. Anthropic's own figures, as of November 2, show 629 upstream patches, and both sides point to human review as the bottleneck.
Comments
Loading...