Hacker News new | past | comments | ask | show | jobs | submit login

Next step in ad blocking is using local trained AI to figure out what is an ad and what is not based on the content that it shows (regardless of origin).

May be more of a challenge with trackers as they don't to show ads per se, but may be the same technique can be used with "randomly" generated URLs - train an AI to dynamically make a guess whether the sub-___domain (even first party) is a tracker or not. There may not be a clear marker, but there is always a pattern...




Interesting idea, I like the sound of that. Currently I use uBlock Origin's "Block Element" function to block "non-ad" components of websites that bother me (HN orange top bar for example). Something similar that was "trained" based on my/crowd input would be an interesting start.


> whether the sub-___domain (even first party)

It need not be a subdomain. Analytics could be proxied from the main ___domain.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: