alt Hacker News"AI safeguards" are not working I guess.. or maybe they're only working against those who'd like to secure their software.. good job Anthropic + OpenAI!
"AI safeguards" are not working I guess.. or maybe they're only working against those who'd like to secure their software.. good job Anthropic + OpenAI!
The AI safeguards are indeed a joke, you can get around their classifier by simply masking out all the unsafe words and it will happily work on your rootkit.