· · AI Alignment Forum
Understanding Safety Filters in AI: Why They Fail and What’s Next
This is the fourth in a series of informal research updates from the Google DeepMind Language Model Interpretability team, in interpretability and adjacent areas. The third post can be found here. Since SFT is the cause for many safety relevant properties, a natural strategy is to filter out rollout
#ai#safety filters#deepmind#technology#india-tech