ASCII Smuggler Tool: Crafting Invisible Text and Decoding Hidden Codes
mediumnewsLLM-Specific
securitysafety
Source: Embrace The RedJanuary 15, 2024
Summary
A researcher discovered that LLMs like ChatGPT can be tricked through prompt injection (hiding malicious instructions in input text) by using invisible Unicode characters from the Tags Unicode Block (a section of the Unicode standard containing special code points). The proof-of-concept demonstrated how invisible instructions embedded in pasted text caused ChatGPT to perform unintended actions, such as generating images with DALL-E.
Classification
Attack Type
Prompt Injection
Attack SophisticationModerate
Impact (CIA+S)
integritysafety
Affected Vendors
OpenAI
Related Issues
Original source: https://embracethered.com/blog/posts/2024/hiding-and-finding-text-with-unicode-tags/
First tracked: February 12, 2026 at 02:20 PM
Classified by LLM (prompt v3) · confidence: 85%