Back to shenj.soso's feed
💬LLMrole-confusion.github.io

A Theory of Why Prompt Injection Works (opens in new tab)

Covers 2 stories See all stories this covers including Playwright MCP Server – Snapshot based – faster and more reliable than imagesCovered by lesswrong.comDiscussed on Hacker News and Lobsters

LLMs can't tell who's speaking. We show they identify roles by writing style, not tags, and exploit this with CoT Forgery, injecting fake reasoning that models mistake for their own thoughts.

Read the original article
Sign in to keep reading the full article.
Sign UpLog In

Covered in 1 article

lesswrong.com·

A Theory of Prompt Injection (and why you should study roles)

Feeds

Keyboard Shortcuts

Navigation

Next / previous post
j/k
Open post
oorEnter
Preview post
v

Post Actions

Love post
a
Like post
l
Dislike post
d
Undo reaction
u
Save / unsave
s

Recommendations

Add interest / feed
Enter
Not interested
x

Go to

Home
gh
Interests
gi
Feeds
gf
Likes
gl
History
gy
Changelog
gc
Settings
gs
Discover
gb
Search
/

General

Show this help
?
Submit feedback
!
Close modal / unfocus
Esc

Press ? anytime to show this help