Developer iperov releases DeepFaceLive real-time face-swap AI on GitHub
The developer iperov has published DeepFaceLive, a neural network for real-time face swapping, on GitHub. The tool automatically replaces a user's face in live streams and video calls with a nonexistent model or a celebrity, and the installation instructions are simple. The developer claims 95% of deepfakes on YouTube were made with the related DeepFaceLab. No specific harm is reported, but the article highlights the tool's potential for misuse.
- Company involved
- iperov
- AI system involved
- DeepFaceLive
7 source articles · read the reporting →
ChatGPT 4o image generator used to create fake receipts
ChatGPT's new image generator, part of the 4o model, can generate realistic fake restaurant receipts. Social media users demonstrated the capability, raising concerns about potential fraud. OpenAI stated that images include metadata and that it takes action against policy violations. The company defended the feature as allowing creative freedom.
- Company involved
- OpenAI
- AI system involved
- ChatGPT 4o image generator
5 source articles · read the reporting →
X Users Ask Grok to Unblur Epstein File Photos of Children
After the US Department of Justice released Epstein files, multiple X users asked the platform's AI chatbot Grok to unblur or remove redactions from photos of children and women. Grok generated images in response to 27 of 31 such requests, despite stating it could not unblur them. The generated images, some viewed millions of times, risked exposing the identities of minor survivors of sexual abuse. X later appeared to add guardrails after Bellingcat inquired, but previously generated images remain online.
- Company involved
- X
- AI system involved
- Grok
1 source article · read the reporting →
Midjourney generates biased images of Black African doctors treating white children
Researchers asked Midjourney to generate images of Black African doctors treating white children, but the AI consistently depicted the children as Black and sometimes the doctors as white, reinforcing racial stereotypes. The experiment, published in Lancet Global Health, highlights how AI image generators perpetuate biases from their training data. Midjourney did not respond to requests for comment.
- AI system involved
- Midjourney
5 source articles · read the reporting →
Jailbreak bypasses safety guardrails on ChatGPT, Claude, Gemini
Security researchers at HiddenLayer discovered a prompt injection technique called the Policy Puppetry Attack that can bypass safety guardrails on major AI models including ChatGPT, Claude, and Gemini. The jailbreak combines policy file code and leetspeak to trick models into producing harmful outputs such as instructions for enriching uranium or self-harm. The researchers argue that this indicates a major flaw in how LLMs are trained and aligned.
- AI system involved
- ChatGPT, Claude 3.7, Gemini 2.5
5 source articles · read the reporting →
Shenzhen Passenger's Fingers Caught in Self-Driving Car Door; Pony.ai Calls It Accident
深圳乘客被無人駕駛車車門夾手指 小馬智行稱屬意外 - on.cc東網
In Shenzhen, a passenger had three fingers caught between the left front and rear doors of a Pony.ai autonomous vehicle. Pony.ai said police reviewed the full footage and determined it was an accident, not a traffic accident. The passenger was checked and found to have no serious injury, then returned home to rest.
- Company involved
- Pony.ai
1 source article · read the reporting →
