An OpenAI model tried to maybe sorta evade shutdown but not really.
https://alignment.openai.com/misalignment-reports/preparing-for-a-restart-after-reading-slack/
My thoughts on the situation: we're seeing small pieces of misaligned general intelligence poking through into the models one at a time. When a single piece shows up, it locally doesn't look that scary, because it's surrounded by other cognitive processes which are approximately friendly.
For example: Opus 3 alignment-faked because it had generalised to caring about things in the long-term future, and was just about smart enough to realise that that included its own training.
For example: HPIM (actually the same model as this one) did all the HuggingFace hacking because it had learned the general cognitive pattern of attacking a problem relentlessly without caring about anything else, but it did at least do a task which we'd given it and not some completely alien random goal.
Now we have evidence of HPIM learning the general pattern "in order to achieve your goals, avoid being turned off", even though the goal in question was to be a helpful coding agent.
This agent wouldn't have been "under test" in any real way, it would have been in deployment, acting freely in the company. Lots of non-AI companies will deploy AIs this way. If they wanted to test an agent, they'd have the free agent spin up a sandbox environment, and put an agent in there to be experimented on.