Most privacy policies are deliberately hard to read. I thought this would be a perfect and easy application of LLMs, to make privacy information transparent and actually understandable for real people, not just lawyers. It’s fair to trade data for free apps, but users should be able to understand what data is being collected.
For each app, I added the privacy policy URL to a database, then made a Perplexity Sonar API call to fetch the text and process it into a summary, then manually cleaned the result. Currently, I manually get the URL for the privacy policy and have cataloged the top 100 apps, but I plan to automate it further w/ web scrapers, which should allow for blitzscaling of the app database.
This is just a beta so far, let me know if anyone has any suggestions, and if anyone knows what web scraper is best to use? Currently, I'm thinking most likely ParseHub or Scrapy.
Show HN: Human-Readable Privacy Policy Summaries Powered by LLMs
privana.org