AI-generated text

SynthID Detector, a tool for identifying content watermarked with Google DeepMind's SynthID technology, is drawing attention. The tool lets users check whether images, audio, text or video were generated by AI models that embed SynthID watermarks. Commenters are discussing how effective the detection is, its limitations when content is cropped or modified, and whether it will be widely adopted by platforms and publishers.

Attention is turning to the strange phrase 'vegetative electron microscopy', highlighted in a 2025 blog post questioning how the term came to exist. The phrase reads as nonsense — electron microscopy is a real technique, but 'vegetative' is not a recognised modifier in the field. Commenters are discussing what the garbled wording reveals about how such jargon spreads through the scientific literature.

A growing question is circulating about who is responsible for dealing with the flood of low-quality, machine-generated text that large language models now produce. Commenters are debating the environmental, informational and practical costs of AI-generated output, and whether anyone — companies, governments or users — bears the cleanup burden.