Not really, especially not a transaction with a high level of ambiguity (mixin). Knowing a transaction originated at an IP address would not let you see where the transaction is going, what the amount is, or anything more than is already revealed in the blockchain (look at any transaction on monerochain.info as an example).
Their point about Tor stands, and it's the same reason why we've
partnered up with the I2P team to implement I2P routing in Monero without reliance on the Java router (and, thus, adding a dependency and an attack surface we can't control - no serious cryptocurrency should allow Java to run on the desktop imho). Premature implementation of I2P would be a bad move, which is why we're taking it slowly.
Thus the only real identifiable risk with using Monero for absolutely untraceable and unlinkable transactions at this stage would be that an attacker with sufficient access to your ISP or your router can determine that you are running a Monero node and that you are relaying transactions that may, or may not, be yours. With sufficiently large control over the entire Monero network (practically impossible for anyone except maybe the NSA) they could possibly determine the origination of a transaction (but again, not where it went or what the amount was). I2P allows us to close that hole.
On that note, though: we are working on a deeper examination of the risks associated with ring signatures and the CryptoNote protocol in general (and how any risks, if they exist, can be mitigated), and will be pushing some more details out as that sub-project progresses.
