I don't think that's true at all. If you're willing to believe that they will do evil things with your chat messages, I don't know why you would have trusted the "This chat is off the record" message in the first place.
I can think of many other possible explanations for the change that, while you may not agree with them, are at least reasonable.
If they were doing genuinely evil things, they'd simply have been lying about not saving logs all along, no?
> Care to share them with us?
If someone has "off the record" mode enabled, you can't send them a message if they're offline. Makes sense, but it reduces the usefulness of the service for both parties and it's kinda confusing.
Who knows. I imagine so. There are 2 aspects -- using it to build a profile on you and collaborating with the government agencies to help them gather information on you, or complying with a wiretap warrant.
They could have still be doing both but secretly. However if you mentioned say "home brew kit" in your chat then started seeing home brew kit ads, it would be a dead give-away of what happened.
As long as they keep XMPP support around for ordinary google talk accounts, you have the power to use clients that support OTR-encrypted IMs. That way, Google only stores useless messages.
Pidgin and bitlbee make this super simple to set up. In fact, by default, my client performs auto-detection of someone else's OTR plugin, which means that after I send my first message to someone, my conversation is automatically "lifted" to an encrypted (albeit untrusted) channel. When set up properly, it's so seamless that I don't even notice it happened.
So there's some problems with re-establishing an encrypted channel. Usually I get a message telling me that bitlbee has to renegotiate the connection, but then I'm back in action.
It seems insecure to leave another point of access open to the conversation, if merely out of principle, due to concerns over leaving a window open for eavesdropping.
There's no reason that has to be insecure (depending on your definition of "secure"). There's also no reason the user shouldn't be allowed to make that choice for himself.
They most definitely do. It would be silly of them not to.
They spent/d real money maintaining it, designing and providing for free.
Unless they are a charitable organization and totally motivated by good intentions the would probably want to scan all the content you pour into their system and build a profile on you so they can sell you better to their real customers -- ad buyers.
