Meta's newly launched tool for spotting AI-generated images failed to identify 55% of pictures made by its own model once they were cropped, according to a Reuters analysis, raising fresh doubts about the reliability of watermark-based detection. The findings, reported on July 10, land just days after Meta unveiled the detector alongside its new image and video generation models.
Continue reading
The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in✓ Signed in — this article isn’t included in your current plan.