Courts have long dealt with this by requiring "real evidence", original documents and first-party testimony.
Traditional news media deas with this by checking sources and by attributing articles to the reporter (or paper). Reporters with a history of "getting it wrong" learned the meaning of unemployment. Nowadays, amateurs can post "news", even anomalously, which has cost us the ability to weed out the un-reputable.
The way to deal with deep-fakes is the same... attribution and fact-checking. There was always someone behind the camera. They just need to be identified so they can earn the reputation they deserve (good or bad).
the bigger problem is in our hyper connected society, even if a deep fake is latter proven to be that, the damage may be done and the tribal truth already decided.
e.g. nancy pelosi slurred speech video, which some trump supporters still holding as fact she was drunk.
and once they are released to the world its difficult to get rid of them, just ask any victim of "revenge" images
if the platform introduces a signature, there are someone will re-create the tools without it.
The only way to tackle it is at the first posting, so it comes down to platforms to have a better AI/ML than was used to stich the video together and tag it as a possible deepfake, and peoples definitions of reputable news organisations to stop including those posting clickbate.