The article highlights Google's previous commitment to watermarking AI-generated content and the need for improved watermark security. It also contrasts Gemini's capabilities with other AI models like Anthropic's Claude and OpenAI's ChatGPT, which refuse to remove watermarks due to ethical considerations. The piece concludes by emphasizing the importance of addressing these issues to protect intellectual property rights and the livelihoods of creatives.
Key takeaways:
- The Gemini 2.0 Flash AI model can remove watermarks from copyright-protected images, raising ethical and legal concerns.
- Google's AI Studio allows watermark removal, but the mobile and desktop versions of Gemini 2.0 Flash do not support this feature.
- Google and other AI companies have pledged to implement watermarking systems to address risks like deepfakes, but Gemini's current capabilities challenge this commitment.
- Removing watermarks from images is illegal and unethical, as it undermines the work of artists and photographers and violates copyright laws.