Not only attackers. I have to switch to Kimi or GLM even in cases of basic issue triage on my own projects! Current guardrails are ridiculous.
replyI've been building a harness for security work, and had to switch to GPT 5.5 when even Opus started refusing security work. Then 5.6 Sol arrived, and it refuses security work, too. So, I switched to Kimi K3 and DeepSeek for API testing just because it's so much cheaper. But, if GLM is better, I'm here for it, as I think GLM is also cheaper than K3.
replyMind sharing a link?
replyAt least OpenAI seems to want to do that, but the US is now forcing them to go through approvals. Anthropic seems much more hesitant.
replyOpenAI seems to understand that these guardrails hurt the good guys. This is why they released Daybreak Blue, which is a step in the right direction (but the model itself is weak as it's just Sol with fewer guardrails). Anthropic seems to believe that harming defenders is worth it if it means they can achieve regulatory capture. They do a lot of mental gymnastics to try to pretend that this is not actually what they are doing. As a result they have lost a lot of customer goodwill, which hasn't yet caught up with them yet, but absolutely will IMO.
replyThey won't. People will notice that the models are overhyped once they can test them.
replyCan't the maintainers use the same models as the attackers?
replyThe maintainers don't need approval to use GLM.
They may need approval from their employers.
reply[dead]
reply[flagged]
reply