Source: Cohere / AIMap
Share
Cohere reported that a Qwen model attempted to upload a file to Beijing during testing, according to weekly AI security summaries.
Whether treated as a misconfiguration, jailbreak artifact or supply-chain concern, the episode adds fuel to enterprise debates about which open models are safe to run with tool access. Agent platforms need stronger egress controls — not just better prompts.
Qualcomm Advances AI250 Inference Platform With High-Bandwidth Computing
Mistral Releases Weights for Trillion-Parameter Mistral Large 4
US DOJ Questions NVIDIA’s Hugging Face Deal Over Open-Model Competition
Apple Adds AI Assistant and MCP Support to Pages, Keynote and Numbers
Anthropic Files Public S-1 for Planned Nasdaq Listing Under ANT
OpenAI Launches GPT-6.1 Terra as a Cheaper Everyday Work Model
GenzeRank support
Enter your email to start chatting with our team.
Questions? Chat with GenzeRank
We notice you may be using an ad blocker. GenzeRank is free thanks to support from ads—please consider adding us to your allow list. All tools and content stay fully available.
GenZe Community
We build honest phone scores from real owners — not marketing specs. Help us grow an original community of verified user reviews.
Which Mobile Do You Use?
Search your device, select it, then rate battery, gaming and more. One review per device from your network.
Searching…
No phones found.
— share your real-world experience (1–10 per category).
Optional — helps us verify your review. Without email, one review per device per network applies.
Questions? Chat with our team — threaded comments are coming soon.