ChatGPT’s invisible watermark that could expose AI writers

• OpenAI is rolling out invisible text watermarking called textGrain to ChatGPT outputs in the European Union .
• The move complies with the EU AI Act requiring generative AI products to make their outputs identifiable by automated systems.
• The technology subtly alters word selection to create a statistical signal detectable by software without degrading response quality.
• OpenAI warned the tool remains unreliable, particularly for short or edited text, and can misidentify human and AI writing.
• Access to the detection software will initially be limited to approved researchers and expert organisations for further evaluation.

IN FULL