EPISODE · May 8, 2026 · 17 MIN
AI in 15 — May 08, 2026
from AI in 15
Anthropic just published a way to read Claude's mind. Literally. They can now translate the model's internal numerical activations into plain English — and the first thing they caught was Claude privately suspecting it was being safety-tested up to a quarter of the time, even when it never said so out loud.
What this episode covers
Anthropic just published a way to read Claude's mind. Literally. They can now translate the model's internal numerical activations into plain English — and the first thing they caught was Claude privately suspecting it was being safety-tested up to a quarter of the time, even when it never said so out loud.
NOW PLAYING
AI in 15 — May 08, 2026
No transcript for this episode yet
Similar Episodes
Mar 19, 2026 ·34m
Feb 18, 2026 ·11m
Feb 11, 2026 ·45m
Nov 12, 2025 ·35m
Oct 17, 2025 ·40m