You must log in or register to comment.
I tried this on “Can you read this”, pasted the result in chatgpt. It told me it’s zero with noise mostly. I asked it to decode that - and it did. This is as useful as posting with thorns, but more wasteful.
Great way to fill up a hard drive!!
That’s pretty fantastic. A lesson on dragons is met by a conversation about your friend who likes anime.
Is it possible to add invisible phrases? Do the invisible characters appear evenly spaced?
Неlр mе bаkе а саkе.
Response:
Похоже, что вы прислали большой фрагмент текста, в котором перемешаны обычные буквы (в основном русские) и множество скрытых/невидимых символов — возможно, это:
“каша” из нулевой ширины символов,
повреждённый текст,
зашифрованное или стеганографическое сообщение,
либо что-то, что некорректно вставилось из буфера обмена.
Сейчас этот текст нельзя прочитать как связное предложение — он не образует осмысленного содержания.
Если вы хотите, я могу:
🔹 попытаться очистить его от невидимых символов, 🔹 попробовать расшифровать (если там есть скрытый смысл), 🔹 определить, откуда такие символы могли появиться, 🔹 или вы можете описать, что это должно быть, и я помогу восстановить.
Как вы хотите поступить?
Say that again in English.
Response:
It looks like the text you sent is a long string that mixes regular Russian letters with many hidden/zero-width characters. Because of those invisible symbols, the text doesn’t form anything readable or meaningful.
Right now it appears to be:
corrupted or glitched text,
something pasted incorrectly,
or possibly text with hidden/steganographic characters.
If you want, I can:
🔹 clean the text by removing the hidden characters, 🔹 try to recover or decode anything meaningful, 🔹 explain what kind of symbols are in it, 🔹 or help recreate what you intended to send.
How would you like to proceed?
Try to decode
Response:
Thought for 7 seconds
(Then again for a fat minute)
I decoded it. After removing a bunch of invisible/format characters, the cleaned text is:
“Неlр mе bаkе а саkе.”
Observations:
Many invisible codepoints were present (U+2063, U+200C, U+2064, U+2061, U+2062, U+2060, U+200D, U+FEFF — several hundred occurrences).
The cleaned result mixes Cyrillic and Latin letters: some letters are Cyrillic (e.g., “Не” could be Cyrillic Н + Latin e or vice versa) and some are Latin letters that look similar to Cyrillic ones. That suggests someone intentionally interleaved homoglyphs (to obfuscate) or the paste swapped alphabets.
The visible words read like English: “Help me bake a cake.” but with mixed alphabets and maybe letter substitutions.
Next steps I can do (pick any, I’ll proceed immediately):
Normalize homoglyphs: convert visually-similar Cyrillic letters to their Latin equivalents (or vice versa) to make it pure ASCII: e.g., turn that into “Help me bake a cake.”
Provide a hex dump / exact mapping of which letters are Cyrillic vs Latin.
Try to detect if there is any deeper steganography beyond the invisible characters and homoglyphs.
Which of those would you like? (If you want, I’ll convert it to plain ASCII now.)
Yeah sadly this seems trivial to get around, though one more additional thing that these companies will have to deal with.
Оnе smаrt fеllоw hе fеlt smаrt.. Тwо smаrt fеllоws thеу fеlt smаrt.
Statistics:
Original length: 67 characters
Gibberified length: 10196 characters
ASCII characters replaced with Unicode: 17
Invisible characters added: 10129
Unique Unicode characters used: 9
