Live data from Hacker News

Tesla has used space characters in internal emails to identify leaks

twitter.com

121–130 of 448 posts

Re: Tesla has used space characters in internal emails to identify leaks

#121

Earlier quoted context omitted.

Corporate networks tend to keep printer logs for extremely long durations. Also, printers at home are incredibly useful. I'm surprised more people don't have them. Especially now that online shopping (and therefore online returns) are so much more common and require printing return labels.

Amazon returns no longer require printing out labels, so someone who does most/all their online ordering via Amazon wouldn't need a printer.

Only for things bought and sold through amazon. If you get something sent directly from a seller returns have you print out a label.

Re: Tesla has used space characters in internal emails to identify leaks

#122
post #88

Earlier quoted context omitted.

The most impressive/unbelievable part of this story is that 2 Word files, created separately, ended up with the same exact file size. I'd assume with all of the Metadata they cram in that this would be unlikely.

Same here, maybe Tesla as a homogeneous word deployment? I’m sure between various version there is difference in what is produced. Unless you can ask word for the equivalent of a postscript ?

I believe the printer logs would contain the size of a Postscript-like equivalent, not the docx itself. I wonder if the story is missing the detail of submitting the new Word file to a printer and checking the size. Or he fired someone random where Word doc size == printer input size.

Re: Tesla has used space characters in internal emails to identify leaks

#123
post #88

Earlier quoted context omitted.

The most impressive/unbelievable part of this story is that 2 Word files, created separately, ended up with the same exact file size. I'd assume with all of the Metadata they cram in that this would be unlikely.

Same here, maybe Tesla as a homogeneous word deployment? I’m sure between various version there is difference in what is produced. Unless you can ask word for the equivalent of a postscript ?

From what I've heard of their engineering practices, it would be shocking if any deployment they have is homogeneous

Re: Tesla has used space characters in internal emails to identify leaks

#124
post #43

Ashlee Vance's biography of Musk (from a few years ago) had a fun similar story. Someone inside Tesla wrote a letter, I believe complaining about product defects, which was printed by a media outlet. Figuring that the leaker printed the letter at Tesla offices (i.e. likely didn't have a printer at home), Musk re-typed the letter in Word, found out exactly how many kilobytes a Word file with that exact text would be,…

To be really secure when leaking typed documents, I get 5 colleagues together and we whisper the leak along the line, so that many of the words are changed by the end. For each word we are randomly spaced to defeat Musk's space-based detection.

We then take the output of this and transcribe it over an old HAM radio to a friend in China. He lives on a farm, so he then dictates it to a chicken who writes it in the sand. He quickly takes a picture before the wind erases it.

He then contacts a US media publisher under the name 'Whu Lee K', and they print it as is! What a wonderful time to be alive.

Re: Tesla has used space characters in internal emails to identify leaks

#125
post #71
post #43

Ashlee Vance's biography of Musk (from a few years ago) had a fun similar story. Someone inside Tesla wrote a letter, I believe complaining about product defects, which was printed by a media outlet. Figuring that the leaker printed the letter at Tesla offices (i.e. likely didn't have a printer at home), Musk re-typed the letter in Word, found out exactly how many kilobytes a Word file with that exact text would be,…

I wonder about this stuff and expect they could also just look for a file of that exact size in bytes on work PCs or cloud storage. I think anyone who wants to write potentially career limiting things should exercise basic common sense like not using work computers and printers, removing fingerprinting elements, etc. In the future I expect more to do stuff like cut and paste between programs to alter the font and spa…

On a modern computer, all copying & pasting between programs would do is just create an exact trail of what is on your computer. Copy -> Paste is not metadata free by any means.

If your final target was Notepad or similar, it should drop all the metadata. But in that case, just start with Notepad.

Re: Tesla has used space characters in internal emails to identify leaks

#126
post #43

Ashlee Vance's biography of Musk (from a few years ago) had a fun similar story. Someone inside Tesla wrote a letter, I believe complaining about product defects, which was printed by a media outlet. Figuring that the leaker printed the letter at Tesla offices (i.e. likely didn't have a printer at home), Musk re-typed the letter in Word, found out exactly how many kilobytes a Word file with that exact text would be,…

This is called parallel construction. The story as told didn't actually happen IMO.

Re: Tesla has used space characters in internal emails to identify leaks

#127
post #85
post #23

Earlier quoted context omitted.

Margaret Thatcher did this in the 80’s to help identify who was leaking cabinet documents to the press. I recall a story about Jobs tracking down leakers in his own executive suite by telling each of them a slightly different things in private. Not sure if it worked, but it can be a very effective deterrent.

I worked with someone who would tell me very precise weird personal details (eg, “the left side of my mouth can’t taste”). I found it odd because they weren’t really pertinent and seemed really unique that I’ve never heard of anyone else with the condition. I suspected he did this to try to determine who they can trust when he would remark that people were liars or not trustworthy and one time a coworker told me a co…

So was he like ... hey bud, don't tell anyone this, but ... "the left side of my mouth can't taste"

Because I can't figure out how this would be helpful in any other context.

Re: Tesla has used space characters in internal emails to identify leaks

#128

What's the best process to strip these hidden characters? Is the an app, or web service that will "clean" the copy?

Translate it back & forth. It'll as a bonus also clear out any specific words used to trap you.

Re: Tesla has used space characters in internal emails to identify leaks

#129
post #61
post #25

Also worth noting that this did not work out, because their general counsel accidentally replied all to the email, giving everyone in the company a safe copy to leak.

"Reply-all" suggests that the same mail went to a set of recipients, but doesn't the watermarking strategy require each recipient to have their own unique copy?

The "to" line is just a header and you can put whatever you want. The actual recipient is specified out of band (RCPT TO). This is how "bcc" works.

Likely the To: line is some sort of mailing list in this case. The mailing list watermarks when redelivering to everybody. The To line remains unchanged. A reply-all causes one copy to be leaked (but it might be re-watermarked, depending on configuration...) Edit: A reply further down says it was actually a forward which does seem more likely, but nothing about the original setup is impossible.

Re: Tesla has used space characters in internal emails to identify leaks

#130
post #23

It’s called whitespace watermarking and seems to be one of those ideas that keeps getting re-invented. It’s used a lot, particularly in HTML where the rendering hides the identifying spaces. It can help you figure out if someone is scrapping your content.

Margaret Thatcher did this in the 80’s to help identify who was leaking cabinet documents to the press. I recall a story about Jobs tracking down leakers in his own executive suite by telling each of them a slightly different things in private. Not sure if it worked, but it can be a very effective deterrent.

The trick is actually very, very old. Mapmakers used to add very small and intentional "mistakes" to their maps so that they could identify who was copying their work (because they would copy the mistakes along with all the legit parts of the map).
Post reply on HN