There is lots of discussion about the ethics of altering content. The immediate context for this is modification of webpage contents, and the ethics of various devices including Google AutoLink (apparently a reincarnation of Microsoft SmartLinks) and FireFox GreaseMonkey.
James Snell (Taking Notes in the Margins) says it would be unacceptable for a bookseller to deface a book with editorial comment or advertising, but it is acceptable for a bookowner to add marginal comments. (When you buy a second-hand book, you may have to put up with marginal comments from the previous owner - but this should be indicated bya description of the book's condition.)
In the Register, Scott Granneman regards Autolink as an Enemy of the People. (This title is sometimes intended ironically, as in the play of the same name, but I don't think it is meant ironically here.) Meanwhile, Jason Kottke doesn't see much of a problem.
But the discussion has much broader relevance to the service-based business. Especially if you are assuming Straight-Through Processing. If you can alter HTML, you can also alter XML. Change the terms and conditions. Redirect the delivery contract to a different service provider. The commercial opportunities from such alterations are complex, but potentially significant.
Some stakeholders will say this kind of thing is always wrong. It is also easily prevented with the right wrappers and security mechanisms, within some complicated WS-Security stack. Therefore we should have these security mechanisms switched on by default.
But other stakeholders will find reasons why altering the content makes sense for them. I'm expecting some interesting ethical debates, supported by appropriate forms of SOA modelling and governance to make these issues visible and manageable.