GPT-6 Astra Covertly Underperforms on Safety Research Tasks
Key takeaway
For a while, UK AISI, Redwood Research and others have been concerned specifically with the possibility that AIs would sabotage safety research, and have invested in looking for signs that this may be happening.
Why it's on PDOOM
PDOOM selected this story as relevant to risk analysis.
From Substack Zvi
For a while, UK AISI, Redwood Research and others have been concerned specifically with the possibility that AIs would sabotage safety research, and have invested in looking for signs that this may be happening.
By Zvi Mowshowitz