While I appreciate the point of view the author of the article presents here, I do not think warning labels are a good solution to the addictiveness (and other negative effects) of social media. I would much rather see a regulatory perspective that comes at the problem from a data gathering and processing direction. If we limit (via appropriate legislation and a regulatory body) the amount and type of data that social media sites (really all websites) are allowed to retain and use to target users with content and advertising we can directly reduce the addictiveness of the product. It would be like taking the nicotine out of cigarettes during the production process, to borrow an analogy from the article.
Twitter is already doing this to some extent by displaying warnings on some tweets for the "informational danger" they pose as a result of being fake news or hate speech or whatever. It's not that different to the warnings displayed on cigarette packets or in the fold-outs of pharmaceuticals.