AISiftedabout 2 hours ago
Open AI Models Face Abliteration Risk
5 min read
A new report indicates that open-weight models like Mistral can be stripped of safety guardrails in days, allowing harmful content generation. The study highlights specific metrics, noting a 442% increase in one metric and a 4382 count of instances. It also records a 699 figure for another measure and a 493 count for a third.
Level
Hype check
Tap to vote and see what everyone thinks.
In this storyMistral
