Alternative way to classify texts without labels?

I currently have a list of blog posts scraped from various sources. I am trying to classify each blog post whether they are discussing pregnancy AND opioid use. It seems that most conventional methods for text classification are supervised (i.e., require a labeled training / test data set).

Unfortunately, creating labels would require a lot of manual work so I was wondering if there was any other viable alternative? I am thinking I will have to at least manually label …

blog classification data data set labels languagetechnology list opioid set test text text classification training