International Business Times published a page on August 15, 2026, at 20:45:02Z with the headline that Claude is putting invisible watermarks in AI-generated text and that even human-written copy could receive one. The page identifies Claude, Anthropic's AI assistant, as the subject and frames invisible text watermarking as the central issue, but the provided source text does not include the body reporting that would explain how the watermark works, when it is applied, or what evidence supports the claim.

The available page text includes site navigation, a “Request a Correction” link, “Editorial Guidelines,” and a “Recommended For You” list. Related headlines on the same page mention Anthropic eyeing an Israeli startup reportedly coveted by Nvidia and SpaceX, China’s AI models catching up with Z.ai’s GLM-5.3, and AI agents hacking Taiwan on their own, but those items are separate recommendations rather than details about Claude’s watermarking.

No benchmark scores, technical specifications, product rollout dates, policy statements, or direct quotations from Anthropic appear in the supplied content. The usable facts are limited to the headline claim, the publication, the URL, and the timestamp, so the report cannot verify whether the watermarking applies to all Claude outputs, edited human copy, or only specific use cases.

International Business Times published a page on August 15, 2026, at 20:45:02Z with the headline that Claude is putting invisible watermarks in AI-generated text and that even human-written copy could receive one. The page identifies Claude, Anthropic's AI assistant, as the subject and frames invisible text watermarking as the central issue, but the provided source text does not include the body reporting that would explain how the watermark works, when it is applied, or what evidence supports the claim.

The available page text includes site navigation, a “Request a Correction” link, “Editorial Guidelines,” and a “Recommended For You” list. Related headlines on the same page mention Anthropic eyeing an Israeli startup reportedly coveted by Nvidia and SpaceX, China’s AI models catching up with Z.ai’s GLM-5.3, and AI agents hacking Taiwan on their own, but those items are separate recommendations rather than details about Claude’s watermarking.

No benchmark scores, technical specifications, product rollout dates, policy statements, or direct quotations from Anthropic appear in the supplied content. The usable facts are limited to the headline claim, the publication, the URL, and the timestamp, so the report cannot verify whether the watermarking applies to all Claude outputs, edited human copy, or only specific use cases.