John Scott-Railton☕ on Nostr: NEW: researchers keep finding we cannot reliably detect speech #deepfakes. We rely on ...
NEW: researchers keep finding we cannot reliably detect speech #deepfakes.
We rely on listening for 'naturalness.'
Problem is, compute power📈& generative models are getting much better at faking just that.
Huge implications for #phishing, #fraud.
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0285333
We rely on listening for 'naturalness.'
Problem is, compute power📈& generative models are getting much better at faking just that.
Huge implications for #phishing, #fraud.
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0285333