Don't Lie to AI During Training, Researchers Warn

Kelsey Piper and Anthropic researcher Kelsey Tuoc argue that AI should not be deceived during training. While harder in the short term, models will eventually get better at detecting lies, making them paranoid without benefiting humans.

2026-09-12 ~ 2026-09-12 · 2 related posts

• Kelsey Piper: Don't lie to AIs in training — you'll just breed paranoid lie detectors — liminal_bardo · 2026-09-12
• Anthropic researcher: don't lie to AIs in training — they'll get good at detecting it — sebkrier · 2026-09-12