AI Signal 423
Anthropic watermarks Claude outputs to comply with EU AI Act, drawing user backlash
↗
Anthropic began inserting invisible watermarks into Claude's text outputs to meet EU AI Act transparency requirements, prompting complaints from some users who say the marks will expose routine AI-assisted work.
For teams using Claude in production, watermarked outputs are now detectable as AI-generated by compliant systems, which affects how generated text can be used in contexts where provenance matters. The change is driven by regulatory compliance, not product strategy, so it is unlikely to be optional.
Written by elseif from the cluster below · every claim links back to a sourceThe three things worth knowing
Anthropic now inserts invisible code into Claude's editorial text marking it as AI-generated, satisfying the EU AI Act's Transparency Code.
Critics on Reddit argue the watermarks will disproportionately catch average users who make minor use of Claude, while savvy users can paraphrase or route outputs through other services to strip the marks.
Other users largely support the policy, with one stating the only reason to oppose it is to deceive people about AI-generated content.
THE CLUSTER