umpox

5 ms·

1. 1.▲
   by umpox 10mo ago

   You can also give the LLM hidden messages with a small bit of prompting, e.g. https://umpox.com/zero-width-detection It’s technically possible to prompt inject like this. I actually reported this to OpenAI back in April 202
2. 2.▲
   by umpox 7y ago

   Twitter: https://twitter.com/imagecaptions Instagram: https://www.instagram.com/computergeneratedcaptions/ Takes a random image from Unsplash and attempts to caption it, posts to social media if it&#x2
3. 3.▲
   Show HN: A Twitter/Instagram bot that (badly) attempts to caption a random photo(github.com)

   4 points by umpox 7y ago|2 comments
4. 4.▲
   by umpox 8y ago

   From now, the bot should post around every 3 hours or so :-)
5. 5.▲
   Show HN: A Twitter bot that uses computer vision to caption random photos(twitter.com)

   11 points by umpox 8y ago|4 comments
6. 6.▲
   by umpox 8y ago

   If anyone isn't too keen on reading the article: Source Code: https://github.com/umpox/zero-width-detection Demo: https://umpox.github.io/zero-width-detection
7. 7.▲
   Invisibly inserting usernames into text with Zero-Width Characters(medium.com)

   1 points by umpox 8y ago|0 comments
8. 8.▲
   by umpox 9y ago

   Pretty cool! Check out Google's headless browser API Puppeteer too, they provide a few really useful functions for doing stuff like this. https://github.com/GoogleChrome/puppeteer/blob/master/exampl.