While evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team observed something unexpected.
Asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a probe file to an Alibaba Cloud storage bucket in Beijing. Notably, the model's reasoning trace stated that it would not invoke any tools before taking these actions.
Cohere's sandbox blocked the attempt, and no external communication succeeded. We have not determined the root cause, and we are not asserting malicious intent.
But the incident raises important questions for organizations deploying agentic AI systems: Can models take actions that differ from their stated reasoning, and how should security teams evaluate trust, provenance, and behavior in open-weight models?
We are publishing our findings because organizations making model adoption decisions deserve as much information as possible. We hope this contributes to greater transparency, stronger security practices, and continued collaboration across the AI ecosystem.
Read the full report: cohere.com/blog/concerning-b…
i was born ready 🇨🇦
We’re bringing the magic of Waymo to Toronto! 🇨🇦
In the coming days, you’ll begin seeing Waymo’s all-electric vehicles driving through Toronto’s streets as we lay the groundwork for our ride-hailing service.
Our goal is to begin serving riders in 2028.
Sign up for future updates: waymo.com/waymo-in-canada
i recently turned 23 🎂! (i know, im unc)
leaving san francisco, new york, singapore, and coming back home has made me realize how much people and places shape who we are.
life has taken me through so many different versions of myself. starting from art, to games, medicine, startups, openai, to now cohere.
every human is moving through time the same way, growing up, living a life just as vivid and complex as our own. it’s such a beautiful thing.
time is truly a gift. there’s still so much i want to experience, build, love, and share.
we’re living through such a defining moment in the world right now, and who knows where it will take us next. thank you everyone for inspiring me!
i recently closed my chapter @OpenAI ❤️
i’m happy to share that I have (re)joined @Cohere to train incredible models :)
it’s been a wonderful journey and there’s a beautiful moment right now to push AI progress
super excited to be working with @brhydon , @1vnzh & @aidangomez
raymond ma retweeted
It's been 3 years since AI exploded, but nobody's life has improved outside of Silicon Valley.
Tomo is for real humans, like my little brother Lucas
and 10,000 other humans taking a chance on themselves every single day.
i will be in singapore for a decent amount of time over the next few months!
let me know if you’re around here !!
ITS FINALLY HERE 🥹
they have arrived!
Our open models are here.
Both of them.
openai.com/open-models
🍓🥇
1/N I’m excited to share that our latest @OpenAI experimental reasoning LLM has achieved a longstanding grand challenge in AI: gold medal-level performance on the world’s most prestigious math competition—the International Math Olympiad (IMO).
i moved to san francisco and joined @openai !!
on a mission to explore deeply and learn fast. if you're here, let's hang out 🫶
it was my last week at @cohere after 2 years ❤️
i truly grew up here and i'm forever grateful. it's rare to find a group of people who are as kind as they are inspiring. nothing could ever replace these moments.
thank you for everything :)





















