Thursday, February 15, 2007

Showing hidden Meta-Information in DOC, PDF and more than 100 other file-formats

Did you hear about hidden information in formats like Microsofts .doc?

We did. Yeah. You too. For most of us this is old news. Read here or here, or ask your favourite big brothersearch-engine.
Everybody should know this - but people everywhere, from government to No Such Agencys keep publishing winword-documents on their websites.

During our penetration tests (and during our internal FileInfo-tests) we came across quite many websites with chatty files, especial .doc. We were fed up to explain this again and again and created a nifty little tool to analyze as many file-formats as possible. If you want to give it a beta-try, check by at Serversniffs "FileInfo". Currently this does ONLY files on webservers, this means the file to be checked has to be on some public webserver. Beware: The check is more than slow and supports only files with a size smaller than 1 MB. It also fails on filenames with blanks or %20. It's BETA. Stuff will get better with our next serverupgrade, which will finally kick SuSe-Linux into /dev/nul.

Examples in Winword, containing a bit of hidden information (and no, we won't post any files with hidden text here!)




It's not only winword that is chatty - we also found loads of PDF-files on websites containing Windows-Usernames of the people who created them. This might get dangerous when you are able to determine the user-structure and naming-convention of an organisation. While many pdfs are clean, there seem to a few PDF-Creator-Tools that we found to be vulnerable by default.

Especially Acrobat Distiller puts realnames or Windows-Usernames into the PDFs Meta-Information: (examples: http://www.verfassungsschutz.de/download/SHOW/symp_2006_abstract_pet.pdf or http://www.nsa.gov/publications/publi00010.pdf, both showing usernames in "Author" and "Creator"-Fields.
This seems to be configurable: Google did a better job, see http://www.google.com/ads/techb2b_news.pdf, while Yahoo puts usernames in many files, like this here http://publisher.yahoo.com/rss/RSS_whitePaper1004.pdf.

Feel free to experiment. FileInfo will display internal Meta-Information for more than 100 File-Formats.

Please drop us a mail you're stumbling over something funny or if you just like the tool- we'll do our best trying to fix stuff or add more file-formats and functionality, and we're waiting for any user-input.

tom

1 comment:

susane said...

My best congratulations for your beautiful site! Perfect !!!

voyance Email