- tl;dr sec
- Posts
- AI, Deepfakes, and Phishing
AI, Deepfakes, and Phishing
A round-up of AI and LLMs being applied to deepfakes and phishing
Soon I’ll write a nice narrative tying everything together and add more links, but for now here’s just my notes.
Deepfake Calls
Finance worker pays out $25 million after video call with deepfake ‘chief financial officer’ - “The elaborate scam saw the worker duped into attending a video call with what he thought were several other members of staff, but all of whom were in fact deepfake recreations” 🤯
“On at least 20 occasions, AI deepfakes had been used to trick facial recognition programs by imitating the people pictured on the identity cards, according to police.”
FCC moves to criminalize most AI-generated robocalls after a deepfake Biden robocall discouraged people from voting
Deepfake Images
Warning: adult stuff here
An underground website called OnlyFake is generating realistic looking photos of fake IDs for just $15. This could streamline everything from bank fraud to laundering stolen funds.
AI-generated explicit photos of Taylor Swift went viral on X. Post-Elon X did a great job of cracking down on it (not).
More than 40 high school girls in New Jersey were similarly victims.
OctoML, the engine that powers a16z funded Civitai, knew that much of the images on Civitai were NSFW (including nonconsensual images of female celebrities and images that could qualify as child pornography), but ultimately decided to keep working with Civitai anyway
Fortunately there are some laws being proposed that would establish civil and criminal penalties for the creation and sharing of nonconsensual explicit deepfake images.