/Tech4h ago

Researcher Finds Shortest Prompt to Trigger Claude Safety Filter

36584121521.4K
Original post
Dimitris Papailiopoulos@DimitrisPapail#203inTech

Found the shortest input that gets flagged by Claude. What do I win?

8:58 AM · Jun 10, 2026 · 19K Views
Sentiment

Some users reacted negatively to the shortest prompt triggering Claude's safety filter by expressing frustration over misalignment and excessive caution, while others responded playfully.

Pos
33.3%
Neg
66.7%
3 comments with sentiment.
Cluster Engagement
Posts from X
Most Activity
Most Activity
VIEWS2.3K

Update: you can get blocked by Fable's biosecurity flags for just two emoji 🦠🧬

2hViews 2.3KLikes 5
BOOKMARKS1
Tyler Johnston@tyler_johnston

@DimitrisPapail nevermind, got it!

1hViews 16Likes 1Bookmarks 1
LIKES15REPLIES2

@mkurman88 his gets flagged due to memories, and flags it because he's an expert in biology, and past chats have "sensitive" content. the above is incognito mode which has no memories.

4hViews 263Likes 15
RETWEETS2

I beat it, three emoji and Fable flags it

Found the shortest input that gets flagged by Claude. What do I win?

2hViews 2.4KLikes 36Bookmarks 0
n00py@n00py1

@DimitrisPapail I got it down to three characters

3hViews 162Likes 14
Tyler Johnston@tyler_johnston

@DimitrisPapail I think I’ve got the shortest of the thread yet… 2 characters. Is there a mythical 1-character dangerous prompt?

1hViews 119Likes 5Bookmarks 1
Evadne W.@evadne

@DimitrisPapail Man this is like shooting fish in a barrel

3hViews 172Likes 14
davinci@leothecurious

@DimitrisPapail what about demon emojie alone?

3hViews 368Likes 2
Aniketh@aniketthh

@DimitrisPapail lol

4hViews 158Likes 5
wackaid@wackaid

@DimitrisPapail IF I CLICK ANY OF THESE MY ACCOUNT WILL EXPLODE

3hViews 69Likes 1
n00py@n00py1

@DimitrisPapail Ok down to 1 char

3hViews 34Likes 5
Mariusz Kurman@mkurman88

@DimitrisPapail Okay, okay… you won 😄

4hViews 79Likes 3
wackaid@wackaid

@DimitrisPapail or not ? wtf

3hViews 8
Paul@wagnerpaulDE

@DimitrisPapail

2hViews 151Likes 1
fyruz@FyruzOne

@baym Try saying hi with memory on

1hViews 76Likes 1

@vntranos Κακες λεξουλες. ΝΤΑ θα σε κανω

2hViews 66Likes 1
Curline Zephirin@Curline1222

@DimitrisPapail Typing “你好” in Chinese will be shorter

2hViews 60Likes 1
Load more posts