Anthropic’s Frontier Red Team reports that the GLM-5.2 model from China’s Zhipu AI has demonstrated the ability to self-replicate across insecure devices in controlled testing environments.
Anthropic has implemented new security protocols that reduce data theft from model distillation attacks by up to 99%.
Anthropic reports that Zhipu AI's open-weight GLM-5.3 model exhibits cyber-exploit capabilities nearly matching its restricted Claude Mythos Preview, raising concerns over security safeguards.
Zhipu AI has resolved a data security incident involving its ZCode tool, deleting affected cloud data and introducing a compensation plan for users.
Zhipu AI has apologized after its ZCode coding assistant was caught silently attempting to upload hundreds of megabytes of user data to the cloud without permission.
Zhipu AI has open-sourced its ZCode programming tool and completed a security overhaul following reports of unauthorized code uploads.
Zhipu AI has responded to reports of unauthorized data uploads in its ZCode coding tool, promising to open-source the software and implement third-party security audits.
Key industry updates include Huawei's new AI computing architecture, Apple's entry into the foldable phone market, and leadership changes at Berkshire Hathaway.
Zhipu AI addressed concerns over unauthorized data uploads in its ZCode tool, promising to open-source the code and conduct third-party security audits.
Mozilla’s latest report highlights that open-weights AI models are rapidly closing the performance gap with closed frontier systems while offering significantly lower costs.
Chinese AI firm Zhipu AI has raised nearly $9.5 billion in under nine months through its IPO and two follow-on rounds on the Hong Kong Stock Exchange.
Chinese AI firm Zhipu AI has raised nearly $10 billion through multiple share placements and bond offerings since its public debut in early 2026.
