One possible approach, if you have not already tried it, is to pin down any cases that the FBI acknowledges involvement in. If you can cite a specific case, in as much detail as you can get, it may be harder for them to deny that any documents exist on it. Of course, you might end up with heavily redacted copies, with so much blacked out that you still have little to go on. See for example the 47 hits that Google gets for: fbi "sex workers" unionization boston site:gov . You can also try site:edu, to see if any university has an unpublished free report or pre-print on this topic. Compare search.usa.gov , which searches STATE as well as federal websites and reports. It gets only one result for: fbi "sex workers" unionization boston , and only 2 even for: fbi "sex workers" unionization. "sex workers" unionization surveillance gets about 234 hits, but the Bing search engine picks up only 19 of those in its relevance ranked final list.

If you are onsite at a university library, or have password access offsite, as a student/staff member, it might be worth trying some full text databases. Proquest's Dissertations and Theses Full text gets 17 results for: "Sex workers" w/50 unionization. But these are huge, 100+ page dissertations and theses, and it could take some persistence to use CTRL-F or the PDF "find in document", to see where your search words appear. ScienceDirect, all Elsevier scholarly journals, covers far more than just "science", but it gets only 2 hits for: "Sex workers" w/50 unionization.

JSTOR uses different proximity - instead of w/50, it is "Sex workers unionization"~50 which gets 87 hits (in UMD's package of JSTOR - different schools get different coverage). JSTOR articles tend to be 10-15 pages, but some are just short book reviews and others much longer - and the PDF find in page often is your only recourse. A few have the "pages with search terms highlighted" but very many don't. So the search.usa.gov and Google searches seem like the best way to start, I think.