[thmstcls]

Random aside to the other commenter's linked articles, I find it a bit coincidental that the supposed "kill switch" environment variable, yolAbejyiejuvnup=Evjtgvsh5okmkAv, decodes from UTF-16LE to UTF-8 as 潹䅬敢祪敩番湶灵䔽橶杴獶㕨歯歭癁 which google translates to "You can't do it without a soul."

[yorwba]

Any even-length alphabetic ASCII string decodes to random Chinese characters in UTF-16LE. Digits and = unlock some Japanese hiragana, Korean hangeul and assorted punctuation, but those only make up a small fraction of the total.

For example, 'backdoor'.encode('ascii').decode('utf_16_le') == '慢正潤牯', which Google Translate turns into "Slow and positive", but it's just nonsense.

[thmstcls]

I'm naive to the translation tech space but is this sort of thing unique to languages like Chinese? I figured all this stuff was mostly solved. Like I wouldn't expect dflhglsdhfgalskjdf to have Google Translate output some grammatically valid Spanish output.

[cvhc]

There is one difference between gibberish Chinese and Latin character sequences. In Chinese, each character indeed carry some meanings (like a word). So I guess the model may hallucinate some output inspired by these meanings. In the case "慢正潤牯" -> "Slow and positive", it actually translated the first two characters literally (慢 -> slow, 正 -> correct/positive/upright).

So equivalent English gibberish would be like "hast prank bibble done anut me me ions." Google translates this one to "对我而言，恶作剧已经完成了。" (To me, the prank has been done.) in Chinese -- very valid sentence, and "¿Me has hecho una broma a mí, Bibble?" in Spanish -- also seems valid.

I guess the model is (over) optimized to generate valid outputs. This can be a feature, so it still translates grammatically invalid but to some degree understandable text (like with typos or non-standard Internet language).

[yorwba]

I think the Latin script might be somewhat protected because random jumbles of letters do appear as serial numbers and whatnot, but for other scripts, anything goes.

I say ҏӲҨЏ ҜъКѠ ЇЩіН гӞэѷ in "Russian", Google Translate says "Let's talk about it".

[simonmysun]

Amazing. How did you find it out?

[thmstcls]

I hadn't looked into that story before so was following the rabbit hole of articles and gists and stuff and saw that some referenced a kill switch via env variable, so I just tossed it into that CyberChef online tool using its "magic mode" and ticked the "intensive mode" box and it was the top result. Just commented because I hadn't seen it elsewhere and figure it might be a little easter egg of sorts.

[simonmysun]

It seems that Google Translate simply output garbage when you input garbage. So it should be indeed a coincidence to get this translation.

[thmstcls]

Wow I didn't realize what implicit trust I put in their translation output. Indeed I just tried some other Chinese -> English translation sites and they vary widely on what they output. Is it gibberish chinese characters these translators just guess on? Either way thanks for the insight I clearly put too much assumed faith in their quality/accuracy.

[cvhc]

Right, completely gibberish. as a native speaker, I can recognize at most 4 characters, and not even one subsequence makes any sense.

Actually just by shuffling these characters you have a good chance to get some specious translations (adding a punctuation makes it more likely to generate a completed sentence): "祪癁番䔽䔽！" -> "I am so sick!" "獶獶祪灵癁癁癁！" -> "The soul is full of blood!"