#opg $OPG @OpenGradient If the verification node does evil, the tokens it has staked are confiscated. This money goes into the system treasury and has nothing to do with the affected users. What makes me even less at ease is this: what if the inference node itself is acting maliciously—giving you a wrong result and then disappearing? Who will be held responsible? The penalty for verification nodes is paid to the network, so you can’t get any compensation; and if the inference node does evil, there isn’t even an entry point for accountability.

This makes me realize something: the incentive mechanism of this network protects the network system itself, not you who uses the network.