That doesn't solve the exploit above | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		hamburglar1 on June 8, 2019 \| parent \| context \| favorite \| on: Grover – A State-of-the-Art Defense Against Neural... That doesn't solve the exploit above

hippiecow on June 8, 2019 [–]

My interpretation is that it would, since if an adversary released a lot of rejection-sampled articles and you retrained Grover on them, then Grover could tag them all as machine-generated.

jasonhansel on June 8, 2019 | [–]

Of course, to retrain Grover, you'd have to already know that they were machine-generated.

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