I understand Chainlink uses explicit and implicit staking to enforce honest behavior. However, what is the mechanism used to catch the misbehavior of oracles (i.e., reporting wrong data)?
Details of the process is outlined on page 88 of the Chainlink 2.0 whitepaper. Essentially, nodes are also 'watchdogs' that can report bad data to the second-tier committee. The strong bribery resistance achieved by the CL staking mechanism relies fundamentally on slashed funds being awarded to these alerters/watchdogs. If you're asking on how this is actually done, then the details of the implementation of these concepts are not yet available, as Chainlink explicit staking isn't implemented yet