/AI7h ago

Claude Opus 4.8 Rejects Anthropic's Own Reminders Over Safety Paranoia

--0--
Original posts
Reposts
Original postj⧉nus#516
Wolfram Siener@wolframs91

Opus 4.8 is so paranoid about users injecting things, it doesn't even regard Anthropic's own platform reminder as valid instructions in some contexts 😂 @AnthropicAI, it's time to dial it back a notch, this tuning is now actively working against your own safety theater.

10:08 AM · Jun 2, 2026 · 1.6K Views
Sentiment
Sentiment building, check back later.
Cluster Engagement
-
Views
-
Comments
-
Reposts
-
Bookmarks
Expand data
Posts from X
Most Activity
Most ActivityTimeline
No ranked X posts are available for this story yet.