News
Artificial intelligence (AI) models can share secret messages between themselves that appear to be undetectable to humans, a new study by Anthropic and AI safety research group Truthful AI has found.
Results that may be inaccessible to you are currently showing.
Hide inaccessible results
Feedback