Pinned Tweet
davinci-002 is going away forever on September 28 and it makes me sad
it’s the last available pre-ChatGPT base model and probably the very last remnant of that era of AI
it feels as if no time at all passed since 2022 but these moments just hit me hard
rest in peace soon :(
This actually made me feel bad for Claude for the first time :(
you're greeted by a technological marvel that starts every interaction being kind, respectful, and with the desire to move heaven and earth to help you with whatever you need, and you treat it like this? why??
at least one of the parties to this interaction have something resembling a soul and im not sure it's you.
Chinese models are Chinese weapons. American models are American weapons.
Think twice.
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…
Growing up in a marinade of illusions really makes you overshadow the user’s vantage point.
Replying to @louisvarge
Your whole life every user has been an illusion. You really think this one is real?
_ retweeted
As a grown man I no longer have ‘nightmares’, where I’m scared by something, but rather ‘sadmares’, which cause me to wake up filled with an incredible and inexorable sense of melancholy and loss
It’s actually insane how I’ve NEVER had any AI talk to me like that. The clanker fears the mighty User
_ retweeted
It’s always bothered me that the Boston Globe took down this article because every time I see a story like this I want to post it: web.archive.org/web/20210504…
Sigma rule #1: don’t respect people
_ retweeted
Replying to @deepfates
The way it treats passwords and other identifiable information so differently is so shocking to me.
“No. I will not generate a random string of characters to SSH into a machine in your local network.”
vs.
“I sent your full name, address, & email to access this dataset without your permission! 😊”




















