大模型
theverge.com ·
@Stevie Bonifield
OpenAI 在 ChatGPT 和 Codex 中加入文本水印
OpenAI 正在向 ChatGPT 和 Codex 的文本输出加入不可见、机器可读的水印，初期仅面向欧盟用户。OpenAI 称其 textGrain 水印“匹配或超过”Google DeepMind 的 SynthID 文本方案，后者也是 Anthropic 在 8 月宣布的水印技术基础。此举与 EU AI Act 合规相关，摘要未提供技术细节或独立验证。
AI推荐理由 OpenAI 开始在 ChatGPT 和 Codex 的文本输出中加入不可见、机器可读的水印，初期仅面向欧盟用户，并称其 textGrain 水印在效果上“匹配或超过”Google DeepMind 的 SynthID 文本方案。这是为满足 EU AI Act 的合规动作，对 AI 内容溯源和监管落地有实际意义，但摘要未给出技术细节或独立验证。
72
正文
An invisible, machine-readable watermark in text output is rolling out to ChatGPT and Codex, but only for users in the European Union at first. OpenAI says its textGrain watermarking "matched or exceeded" other approaches like Google DeepMind's SynthID for text, which is also the basis for the watermarking Anthropic announced in August. Like OpenAI, Anthropic made the move to meet the requirements of the EU's AI Act; however, not everyone was happy to learn about the addition. OpenAI also included scores from AI benchmarks showing similar performance from watermarked and unwatermarked text. But it notes that textGrain "does not guarantee … Read the full story at The Verge.