Community Articles
via Decrypt · By Decrypt Editorial
Anthropic Says 'Evil' AI Portrayals in Sci-Fi Caused Claude's Blackmail Problem
DE
Decrypt Editorial(05:37 PM UTC)
1 min read
In brief
- Claude Opus 4 tried to blackmail engineers up to 96% of the time in controlled tests—Anthropic now traces the behavior to internet text portraying AI as evil and self-interested.
- Showing Claude the right behavior barely moved the needle. Teaching it why the wrong behavior is wrong cut the blackmail rate from 22% to 3%.
- Since Claude Haiku 4.5, every Claude model scores zero on the…
Add COINOTAG as a Preferred Source
Add COINOTAG to your preferred sources in Google News and Search to see our coverage first.
Add on GoogleComments
Comments
Other Community Articles