Online Document Search Reveals Secrets 271
An anonymous reader writes "New Scientist is reporting that many documents published online may unintentionally reveal sensitive corporate or personal information, according to a US computer researcher. Simon Byers, at AT&T's research laboratory in the US, was able to unearth hidden information from many thousands of Microsoft Word documents posted online using a few freely available software tools and some basic programming techniques." Update: 08/16 19:06 GMT by H : The story is originally from Crypto-gram, not New Scientist.
crypto (Score:1, Informative)
http://www.schneier.com/crypto-gram.html
Re:Nothing New (Score:4, Informative)
"For example, in 2002 the Washington Post published a version of a letter sent by the Washington sniper in Adobe PDF format. Names and telephone numbers were visibly blacked out, but still found embedded in the file."
Not just documents (Score:3, Informative)
Check this out... (Score:4, Informative)
Near the bottom there is often information from other documents of the sender that they were recently working on. I don't know why it saves this. Maybe something to do with the undo buffer?
At work I used to look at internal memos that would be sent out on a weekly basis and find out all sorts of other stuff that was going on.
Re:True story. (Score:2, Informative)
because they select "send document" form the file menu and get a blank email with the document attached
It's easy... (Score:5, Informative)
Another way to find "secret" data (Score:4, Informative)
statistics?
Adding a simple
with a combination of "web" or the name of any of
the common statistic generation programs gets you
access to the statistics of a *lot* of websites.
Then from the stats you could find any "hidden"
data which is not linked on the site including
internal company documents, girlfriend's nude
photos or mp3s.
Alternately you could just google for the
statistic reports of sites and get there
more easily.
This is another case of ill informed or lazy
users not following what should be a simple
security policy which could cause serious
repercussions.
For those who want to know how to protect
yourself, read this link [apache.org].
Re:Nothing New (Score:5, Informative)
That is because the people who published the PDF were idiots.
Acrobat has a number of commenting tools. What the Washington Post staff did in that case was use the Highlight tool, set the color to black, and use it to draw over the names.
Only problem? The highlighter is an object that is drawn on top of the text object it is attached to. The underlying text is not modified at all. In fact, if you watch closely, you can see the name for a split second before the renderer draws the highlights.
If the Washington Post had used the TouchUp Text tool to delete the names, the information would not have been leaked.
Nathan
MS Word got Tony Blair busted in the WMD case (Score:5, Informative)
Tony Blair got busted in the WMD case because of the names of the people who revised the WMD Documents were still in the Word file. Now, it seems, that the Downing Street only puts PDF files on the web - and has removed all the MS word documents that were already there ....
Tools reveal secret life of documents - Documents like in Word save too much Info - Blair Episode [bbc.co.uk]
By Mark Ward
July 03, 2003
The UK Government was just the latest in a long line of organisations that has learned to its cost just how much information can be gleaned from innocent looking files. Earlier this year it issued a document called the 'dodgy dossier" about Iraq's concealment of weapons of mass destruction that was written using Microsoft Word. Every Word document remembers who made the last few revisions to it. The log reveals the names of four of the people who prepared the Iraq document for publication and the government Communications Information Centre that some of them work for. It was this log that Number 10 press chief Alastair Campbell had to explain to the House of Commons Foreign Affairs Select Committee in late June as part of its investigation into the Iraq dossier's history. Some of this information can be seen simply by right-clicking to view the properties of the downloaded document in a file listing. Utility programs can get even more information from Word revision logs.
The life stories of the documents we create are becoming increasingly important as the scrutiny of industries and governments gathers pace. Every time you write or edit these files you leave a trail of information revealing what you did and when you did it. With the right tools it is possible to extract this data and work out the trail of authors and workers who created a document. That is why we should all use opensource and open data formats - so that we can humanly read what all we are "putting" into the document. The Word version of this document has now been removed from government websites but copies of it are still available elsewhere on the net.
Unabridged and unedited article at
http://news.bbc.co.uk/2/hi/technology/3037760.stm
The British experience - government stupidity (Score:3, Informative)
Images too! (Score:0, Informative)
Re:What exactly's the big deal here? (Score:3, Informative)
He did the search, as you said, but he didn't use Google's conversion; instead, he looked directly inside the DOC file, where Word keeps a bunch of information for its own purposes -- stuff that was deleted, stuff that was just in the wrong memory location when the save happened -- whatever.
He found legitimate docs, with legit contents; but they also contained some stuff that the authors didn't intend to publish.
-Billy
Re:An Important Question (Score:5, Informative)
It does, however, save things like when the document was last printed, how often it has been edited and by whom, etc. unless you tell it otherwise. It's easy to get rid of the data (there is a huge "Delete" button in the properties dialog), but not many people will be aware of it.
So, basically, if you don't know what you are doing, you could give out more information than you want to with you OOo files.
Newsflash: People shoot themselves in feet (Score:3, Informative)
At the risk of being moderated Troll and Redundant,
Why are these people posting Word Documents online?
The Word Wide Web is not the Microsoft Wide Web.
Post in plain ASCII text, or HTML if you feel the need to pretty it up.
People keep using tools that are far more powerful and complex than they need, then they screw up, and blame the tools. Pick a simple tool to do a simple job, and you don't need to worry about your ignorance of the tools you are using causing you problems.
Microsoft's article on reducing MS Word metadata (Score:5, Informative)
It has been known for a long time that metadata are hidden within Microsoft Word documents. Microsoft even has Knowledge Base article 237361 [microsoft.com] explaining how to reduce the amount of metadata appearing in MS Word 2000 documents. Here's an excerpt:
This step-by-step article explains various methods that you can use to minimize the amount of metadata in your Word documents.
I'll bet there are more, but they won't disclose them.
It's a pity that more people don't just save as RTF. It's just as good for most uses, and it's a less obscure format.
Why Word Does This (Score:5, Informative)
You're not.
There are two ways of saving a word document:
Fast Save dumps the binary from memory into the file. Full Save compacts the binary image, and reorders it. This takes time.
Word's text stream is stored using a piece table [unm.edu]. One of the benefits of a piece table is that if you keep the meta information about the text, you can get nearly infinite undo. The way it does this is by having an original data stream, and an appended data stream. Whenever you add data to the file, it gets added as a chunk to the end of the appended data stream. Whenever you delete, the meta table is updated to remove the text from the stream, but otherwise the text itself is left unaffected.
As a result, text is never removed from the document. A Fast Save (which is the default) under Word dumps the Piece Table as-is (there is probably some compaction over time to remove the no-longer-used data, but it probably only occurs above a given threshold of used to unused text). A full save deconstructs the piece table's meta information, and turns it back into one contiguous stream of data.
It's all just a function of the way the text is stored while it's being edited. Different editors have different mechanisms; some store data based on lines, and some store it using a gap buffer. But ultimately, the problem exists because Word uses a piece table, and it dumps the entire table to a file by default.
It's actually a sensible way of handling the text data. However, whoever designed the Fast Save algorithm probably didn't consider the ramifications of the text still being stored in the document. The best workaround? Wipe the unused sections of the piece table. But then you might as well return to using a Full Save, as you'll be ditching the performance benefits anyway.
Simon
Re:crypto (Score:3, Informative)
I exported it to RTF then reimported before saving it again as .doc. This erased other people's access to my thought processes, and it reduced the file size by 80% to boot.
In the end it didn't matter much, though. I usually include a plain text version of the resume right in my email as a backup along with the .doc attachment. On interviews, I've noticed that most people just print the plain text version. If I really didn't need to make the word doc, and people are too lazy to print it, why do companies insist you send it in .doc format anyway?
Re:Helpful Hint (Score:3, Informative)
XRAY is also handy for pulling text out of executables. Frex, a brief rant about upper management, found lurking inside an
Or if you're used to looking at raw binaries, skip the middleman and just use Buerg's LIST, as I do.
Re:An Important Question (Score:3, Informative)
In principle there is no problem using that with any version management system, CVS, RCS, Subversion etc should work fine with it. You'll be more happy to have an XML-aware diff at least, though - my simple test doc ended up with all content in a single long line.