What you see is what you sign (?)

István Zsolt BERTA

istvan.berta@microsec.hu

Microsec Ltd.

Read the document before signing it. It might be straightforward to follow this principle in the paper based world, but it is much more difficult to do so in the electronic one. An electronic document is represented as a series of bits. When the electronic signature is computed, these bits are hashed and the resulting hash value is encoded with the signatory's private key. The signatory is a human, a natural person, who would like to make sure that the document to be signed is the same as the document on her screen.

Before the bits representing a document are signed, they go through a series of transformations, some information is appended (e.g. a reference to the algorithm used of signing, the X.509 certificate of the signatory, a reference to the signature policy in use, etc.), and the hash that is actually signed is generally not the hash of the document, but the hash of a standardized (e.g. ETSI TS 101 903 or PKCS#7) structure. If we would like to view what we are going to sign, it is not clear which part of the process we need to examine. Perhaps, we would see just a series of bits, and a natural person would not be able to decide if these bits correspond to the meaningful document she would like to sign.

Signature creation applications generally display the meaningful document before signing it. Most document formats we use are complex (e.g. Word, Excel, PDF, HTML) and are very hard to display unambiguously. It is also hard to decide if the person who shall verify the signature and view the document will see the same document. Different parties may use applications of different types with different versions, settings, localizations and environments for viewing the document, and thus make the document display differently. Some document formats also allow so-called active contents (like macros) so it is possible to create documents that display differently in different environments or at different points of time.

We should still read the document before signing it, but this does not necessarily mean that we want to see what we sign; in fact the meaningful document is what we would like to view before we sign it. In our paper we examine the above problem.