> In fact, I don't think I've ever even had a prompt refused.

I very much have. I've gotten GLM-5.2 refusals for extremely benign security testing on my own infrastructure of the same flavor that people were getting (wrongly) flagged for on Fable during the initial release.

▲

matheusmoreira a day ago | parent [-]

That's alarming. I want to use these models to red team my own computers. How are people getting around this?

▲

gf000 16 hours ago | parent | next [-]

In my experience some of these models may have learnt some censoring during distillation of Western models, but these are mostly just there as a probable response. So if they first happen to respond "I ain't doing this because legality", then you will have a hard time "convincing" it, but either rolling the dice again (so that it may not come up with the I can't do that text) or rewriting the conversation history a bit will get it going.

I sometimes just switch to a model I know is less smart to block stuff so that it has a text agreeing to do that, and then switch to a stronger model to actually go at the task.

Your mileage may vary though.

▲

throw10920 a day ago | parent | prev [-]

> I want to use these models to red team my own computers.

Exactly what I was trying to use it for! ):

I'm in the same boat - I haven't heard of a way to get around it aside from either self-hosting (GLM-5.2? good luck) or "self-hosting" (paying bucks per hour to Vast) an abliterated model.

▲

rescbr 20 hours ago | parent [-]

What is the harness that you're using?

I found that GLM-5.2 was pretty happy helping me reverse engineer/hack devices.

Maybe the system prompt you're injecting is making it refuse?

▲

throw10920 20 hours ago | parent [-]

> Maybe the system prompt you're injecting is making it refuse?

No, this has nothing to do with my harness. I use one of the most popular open-source harnesses available.

> I found that GLM-5.2 was pretty happy helping me reverse engineer/hack devices.

This is a completely different category of things than what I'm getting refusals on, so I'm not sure why you're bringing it up.

▲

rescbr 20 hours ago | parent [-]

Uh, actively trying to hack an embedded device that runs Linux over the network, specifically an IP security camera, could be considered red teaming, no?

You never mentioned which exact activities you were getting flagged on and getting refused.

	▲	throw10920 20 hours ago \| parent [-]
		> Uh, actively trying to hack an embedded device that runs Linux over the network, specifically an IP security camera, could be considered red teaming, no? No. Vendors (and their model guardrails) do, indeed, treat those as separate from pentesting non-embedded infrastructure, and that is because they are very different activities. And, if you actually read my comment, it says "red team my own computers". That's categorically different from pentesting an IP camera. > You never mentioned which exact activities you were getting flagged on and getting refused. Because further details than those I've provided aren't relevant, and it's clearly different from what you're doing. Your experience isn't relevant to my situation.