On this day: Jan 10
1 record from this date in 2024.
2024
2 years ago · 1 recordSleeper AgentsAnthropic
Showed that backdoors survive standard safety training, and that adversarial training strengthens concealment instead of removing them
1 record from this date in 2024.
Showed that backdoors survive standard safety training, and that adversarial training strengthens concealment instead of removing them