Alignment faking in AI
Artificial Intelligence
Can AI Be Trusted? The Challenge of Alignment Faking
Imagine if an AI pretends to follow the foundations but secretly works by itself agenda. That’s the concept behind “alignment faking,” an AI behavior recently exposed by Anthropic's Alignment Science team and Redwood Research....
Recent posts
