New tool strips watermarks from AI-generated images

A new open-source tool, `remove-ai-watermarks`, has been released to strip visible and invisible watermarks from AI-generated images. The tool targets watermarks and metadata embedded by major AI image generators including Google Gemini, DALL-E, Stable Diffusion, and Midjourney. It employs techniques like reverse alpha blending for visible logos and diffusion-based regeneration for imperceptible watermarks, alongside metadata stripping and an 'analog humanizer' to bypass AI detection. AI

IMPACT Enables users to bypass AI detection and "Made with AI" labels on social platforms, potentially impacting content authenticity and platform policies.

RANK_REASON The cluster describes a new software tool that manipulates AI-generated content, rather than a core AI model release or research.

Read on Hacker News — AI stories ≥50 points →

• Adobe Firefly
• C2PA Content Credentials
• DALL-E
• Facebook
• Google Gemini
• Instagram
• Midjourney
• remove-ai-watermarks
• Stable Diffusion
• SynthID
• X (Twitter)

• product
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →