Bitcoin nodes consider the chain with the most work the best chain. Whenever a miner finds a new block that extends the best chain, they broadcast it to the network and all nodes update their chaintip as they hear about it. Miners could ignore that new blocks are found, but this would not affect any other network participants.
They could however instead provide a competing chaintip. As your scenario states that the attackers have a majority of the hashrate, (among other things), they could build a chain of empty blocks, i.e. blocks that only pay the block subsidy to the miners, but don’t confirm any transactions. This would allow the attackers to collect the block subsidy, but they would no longer be collecting transaction fees, reducing their revenue by (currently) about 15%. However, the delay in confirmations would drive up the feerates of the waiting transactions, and the lost revenue would quickly increase.
Note that the attackers must continue to mine blocks in order to maintain their denial of service, or they would be out-competed by the minority eventually. So, the attacker is still spending energy cost (and has already invested in hardware). Depending on how the scenario exactly specifies “no longer bound by financial incentives”, I’d surmise that the rising transaction fees would eventually entice miners to break rank from the attack.