Anthropic extends Claude text watermarking to older models
11/05/2026 — 11/05, 03:17·1 sources (all)·1 reports (all)
Key takeaways
- Anthropic told API customers that from November 16 it will add text watermarking to Claude Opus 4.6, Opus 4.5, Sonnet 4.5 and Haiku 4.5, ahe…
Background & analysis
According to The Decoder, Anthropic has emailed API customers again about a second round of text watermark retrofits. Starting November 16, 2026, Claude Opus 4.6, Claude Opus 4.5, Claude Sonnet 4.5, and Claude Haiku 4.5 will all get text watermarking, ahead of the EU AI Act's December deadline. On Amazon Bedrock, the rollout may take a few extra days.
The watermarks only take effect when temperature, top_p, and top_k all sit at their default values. API users who change a single sampling parameter can switch the watermark off on these models. Asked about this gap, Anthropic replied with a compliance warning rather than a fix.
Two details in the material do not line up, and the coverage does not explain why. The body states that Claude Opus 4.6 was originally part of the first wave and should have received the watermark by a date the text cuts off at, while the summary names Claude Opus 4.6 and Sonnet 4.6 as the older models where the watermark is easy to bypass. The summary's Sonnet 4.6 never appears in the new batch list, which instead includes Claude Sonnet 4.5 and Claude Haiku 4.5.
As of the available reporting, the timeline stands at this: Anthropic has sent the notice, the December 16 cutoff for the EU AI Act has not yet arrived, and no information about how the watermark performs in practice after launch has been reported.
AI-generated from 1 reports · updated 22 hours ago
Related AI tools
- Heat index
- 53
- All sources
- 1
- All reports
- 1
- First seen
- 22 hours ago
Reports on this story headlines open the original
Anthropic has emailed API customers that Claude Opus 4.6, Opus 4.5, Sonnet 4.5, and Haiku 4.5 will get text watermarking starting November 16, ahead of the EU AI Act's December deadline, The Decoder reports. On these models the watermark only applies when temperature, top_p, and top_k are left at their defaults, and Anthropic's response to the loophole is a compliance warning.
The DecoderAI score 53
Other stories people are talking about
- Heat index 338RisingAnthropic open-sources FlashProve to prove Claude responses7 sources
- Heat index 78Developer benchmarks Claude Code with and without frontend-design skill1 sources
- Heat index 45Synthesia 宣布可在 ChatGPT 和 Claude 对话中直接生成视频1 sources
- Heat index 136Claude cracks the 3D structure of Vault particles, Anthropic says2 sources
- Heat index 89NewICML 2025 Paper Advises Against Same-Family Model Code Review1 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion