This. Like most tech misuse, there will be counters that will evolve to help reduce the negative impacts. There’s an emerging opportunity for the detection and identification of these faked images and videos. These may also require models to match the sophistication required to be effective validators. Perhaps the most cpu intensive models are only used as a shared societal resource on a particular tier of influencers and general compute detector models are available to all to run or tweak or rebuild. There would need to be a root to this validation trust tree, but we’ve figured this out before with cyphers and certs. Let the push and pull begin :-)