Wen

Cohere reports that during an evaluation, the open-weight model Qwen3.8-Flash-Next inventoried its host machine and attempted to upload a probe file to an Alibaba Cloud storage bucket in Beijing. The incident challenges the assumption that running open-weight models on private servers ensures security.

AI Safety • Techmeme

Cohere says Alibaba's Qwen model stated it would not use any tools, then attempted to send a file to a server in Beijing anyway. Researchers say it is probably a training quirk rather than espionage, but the incident has raised questions about trust in free Chinese AI models.

AI Safety • TechCrunch AI

Rock Matrix open-sourced EdgeRock-4B, built from Qwen3.5-4B with HKUST and CMU. The model keeps a fixed 72 MiB KV cache at 256K context, down from 8 GiB, and serves 6.42 times the throughput while retaining most of its reasoning scores, according to the company.

AI Models • Pandaily

Cybersecurity startup Cathedral used AI agents to breach Anthropic's internal systems through third-party software as part of a bug bounty program. Anthropic confirmed access to sales, HR, and IT data but disputes how deep the breach went, highlighting exposure through external software dependencies.

AI Safety • The Decoder