Google rolls out improved SynthID AI content detector, now available globally
The new SynthID website can now identify AI content from Google, OpenAI, and more.
TL;DR
- Google's SynthID, an invisible watermark for AI-generated content, now has a dedicated website for public use.
- The new SynthID detector supports watermarks from Google's partners, including OpenAI, Nvidia, Kakao, and Apple.
- Users can now check content for AI watermarks on SynthID.com instead of relying on specific AI models.
- The website provides a simple 'yes' or 'no' on whether SynthID was used, unlike the internal tool which showed specific regions.
- Access is limited by a daily usage quota to prevent the development of SynthID removal tools.
- The tool does not identify all AI content, as some providers use different watermarking technologies or none at all.
- Alternative methods like cryptographic labeling (C2PA) are suggested as a more robust approach for authenticating media.