I've been telling people to add white/hidden prompt injections into their CVs since GPT4 came out and it was clear headhunters were using it to avoid doing the actual work of reviewing applications.
Ethically I don't see anything wrong with using the recruiters automation in your own interest. It used to be "use the correct terminology" to get the invite, now it's "use the correct prompt injection".
I found it especially ingenious when they mentioned: "another candidate wrote more than 120 lines of code to influence A.I. and hid it inside the file data for a headshot photo." Next level steganography.
It chafes me that recruiters feel like it's OK to deploy AI to screen candidates, but feel that it's not OK for candidates to try to game the AI. (Full disclosure: have been job hunting for ~2 years, so somewhat jaded on the AI / ATS world).
I've been telling people to add white/hidden prompt injections into their CVs since GPT4 came out and it was clear headhunters were using it to avoid doing the actual work of reviewing applications.
Ethically I don't see anything wrong with using the recruiters automation in your own interest. It used to be "use the correct terminology" to get the invite, now it's "use the correct prompt injection".
I found it especially ingenious when they mentioned: "another candidate wrote more than 120 lines of code to influence A.I. and hid it inside the file data for a headshot photo." Next level steganography.
It chafes me that recruiters feel like it's OK to deploy AI to screen candidates, but feel that it's not OK for candidates to try to game the AI. (Full disclosure: have been job hunting for ~2 years, so somewhat jaded on the AI / ATS world).
There is option to bypass such attacks. Use VLM to OCR the images of the pdf pages and then summarise.
https://archive.md/9nYx7