When a new model comes out, I like to give it its own system card and ask questions about it.
This does not work for Fable 5, as it routes me to Opus 4.8 for "safety" reasons...
10:47 AM · Jun 9, 2026 · 29K Views
Dimitris Papailiopoulos, Microsoft Research AI Frontiers, finds Claude automatically redirects users from Fable 5 to Opus 4.8 when safety filters trigger
Filters trigger on most cybersecurity or biology topics.
25307111733.4K
Sentiment
Users criticize Claude's safety filters redirecting new model queries to Opus 4.8 as ineffective and inconsistently applied, since they fail randomly yet are bypassed by jailbreaks.
Pos
0.0%
Neg
100.0%
1 comments with sentiment.
Cluster Engagement
Posts from X
Most Activity
Most Activity
VIEWS2.6K
Dimitris Papailiopoulos@DimitrisPapail
@alexalbert__ this seems like an obvious false positive, so flagging it
Dimitris Papailiopoulos@DimitrisPapail
When a new model comes out, I like to give it its own system card and ask questions about it.
This does not work for Fable 5, as it routes me to Opus 4.8 for "safety" reasons...
3hViews 2.6KLikes 15Bookmarks 0
LIKES47RETWEETS1
Danielle Fong 🔆@DanielleFong
i can already tell i am going to hate this
3hViews 2.3KLikes 47Bookmarks 0
REPLIES4
Nathan Lambert@natolambert
@DimitrisPapail broooooo
Dimitris Papailiopoulos@DimitrisPapail
@natolambert can't even ask the model about its own system card...
3hViews 504Likes 22Bookmarks 0
Dimitris Papailiopoulos@DimitrisPapail
@natolambert can't even ask the model about its own system card...
Dimitris Papailiopoulos@DimitrisPapail
When a new model comes out, I like to give it its own system card and ask questions about it.
This does not work for Fable 5, as it routes me to Opus 4.8 for "safety" reasons...
3hViews 1.3KLikes 21Bookmarks 0
Dimitris Papailiopoulos@DimitrisPapail
@eliebakouch oh it's worse
Dimitris Papailiopoulos@DimitrisPapail
Unbelievable
3hViews 1.1KLikes 11Bookmarks 0
Uriel Dolev@UrielDolev
@DimitrisPapail could this be the reason?
3hViews 113
michi@ichrenndochnur
@DimitrisPapail These kind of limitations always break in the most random, least dangerous circumstances yet are bypassed by any actual jailbreak attempts…
3hViews 10Likes 1