dailyai.report

A new analysis demonstrates that SynthID watermarks can be removed or altered through simple image manipulations. These techniques bypass Google's attempt to identify AI-generated content. The vulnerability suggests that current digital watermarking cannot reliably prevent deepfake proliferation. Practitioners should treat these watermarks as fragile signals rather than absolute proofs of origin.