EmbeddingGemma 2: an open, lightweight multimodal embedding model
deepmind.google·6d ago
AI Engineering
Hacker News·Sep 10·5 min read
Proof of Capture: Apple Reference Image, but open source and using steganography
TL;DR
A new open-source camera project uses steganography to embed a cryptographic signature in images, proving their authenticity at capture time. This approach counters the growing challenge of AI-generated images by ensuring that real photos can be verified.
✦ Why It Matters
Consider building your own Proof of Capture camera to enhance image authenticity verification.
Key Takeaways
- Explore building the open-source camera for under $100.
- Utilize steganography for embedding cryptographic signatures in images.
- Understand the limitations of current image verification methods.
- Stay informed about advancements in image authenticity technologies.
How It Works
The camera embeds a signed perceptual hash into the image pixels using steganography, ensuring the signature survives edits and compression. The ATECC608 chip securely manages the cryptographic keys, allowing for verification without exposing sensitive information.
⚠ The Catch
The system does not prevent screen attacks, where a photo of a screen displaying an AI-generated image can still be signed as authentic.