More ETH/= more power. In my understanding, an attacker would need to control half of all staked ETH to stage a 51% attack, which would require many billions of dollars. But actually exercising this control by attacking the network would undermine the network’s security/utility, potentially sending ETH to zero and obliterating the wealth of the attacker.
This is, supposedly, a strength of PoS: the more ETH you have, the less incentive you have to attack/destroy the system.
In such a situation, the price effect of the negative news from the attack would be partially canceled out by the jarring reduction in ETH supply.
Trying to cheat through the protocol is uninteresting. You don't destroy the system by breaking L1 rules, any such attack if successful would just get rolled back by hardfork (and nevermind the 'staked money is lost if cheating is detected' system, no one would willingly trigger it).
More ETH is automatically equal to more power, because that's exactly what money is defined to be. We trade it as power.
With proof of work, the same effect exists, thought it involves a loop where hashrate is exchanged for money, which is converted into more efficient hashrate at a TSMC fab.
An actor that openly attacks the network will quickly be blacklisted in either system, but when you control a majority of the money there are other ways to influence the system that don't involve breaking a formal rule.
Disclaimer: I don't have the faintest clue what I'm talking about. I only have passing familiarity with cryptocurrencies, and I've read one or two of their whitepapers.