@kyletut
AI fraud will extend well beyond the restaurant receipt example that is trending on LinkedIN.
Yesterday, I asked ChatGPT to generate a fake video dataset with timestamps, device ids, face embeddings, and more.
From there, I asked it to generate an image that matches one of the rows in the dataset.
While there is still a ways to go on the image/video replication, the dataset is already a big enough attack vector.
Is your data even your data?
0 reply
1 recast
39 reactions