Machine translation on the Web.

I frequently see claims about how helpful the tools that translate Web pages from one language to another are, especially for those of us who are monolingual. Here's a typical claim from Google:

One of our objectives in Google is to supply the access to the the whole world information. A great obstacle for the this is the language barrier. If the ideal page of the result to a question will be written in a language that you do not understand, to follow above until now he will be very hard to start the access to this information. Today, we launch a new characteristic in Google we translate that he makes great one-step examination to direct itself to this problem...

When the translation of machine will not be perfect, is generally good sufficient you to get gist of the information in a language that you could in another way to be incapable to reach. We think that this characteristic will be particularly useful for our international users since even so the majority of users of the Internet for is there either English speakers non, a majority of the index in the Internet we are still in English.

Well, I admit that's not exactly how they put it on their English-language website. It's the result of using Google's service to translate two of their paragraphs from English to Portuguese, and then, since I don't understand Portuguese, translate the result to English. So we should probably assume that only half the errors and infelicities are due to translation in either direction. Nonetheless, it raises two important questions for me:

• The right to make a translation is one of the rights associated with copyright. How much of a website can a commercial site translate for its own profit (remember, it's all about ads) and still claim it as fair use?
• Some people actually rely on the accuracy of information on the Web--especially if it comes from an apparently authoritative source--and base decisions on it. Suppose an error in translating a company's website led to a commercially important miscommunication? Who would be liable? The company that published the Web page? The company that (mis)translated it? Or the user who believed the mistranslation? Similar questions arise about libel. Is this just one of those things that nobody in particular is responsible for, an "Act of God"?