Has Google’s AI watermarking system been reverse-engineered?

A developer known as Aloshdenny claims to have reverse-engineered Google DeepMind's SynthID watermarking system, demonstrating how to extract watermark mechanics from AI-generated images using signal processing and sample data. Google disputes the claim, stating that SynthID remains robust and hasn't been systematically broken. The incident highlights potential vulnerabilities in AI content attribution systems and raises questions about watermarking as a safeguard against synthetic media misuse.