产品多源确认
OpenAI 将在欧盟为 ChatGPT 和 Codex 文本加隐形水印
精选理由
OpenAI 要在欧盟给 ChatGPT 和 Codex 的输出加隐形水印了，检测工具还不对外公开，想绕过的话同义词替换就行。
OpenAI 宣布将在数周内依据欧盟 AI 法案，为 ChatGPT 和 Codex 输出的文本嵌入隐形水印。公司称测试中水印未影响模型能力、速度和文本可读性。用同义词替换可明显降低检测率，短文本的检测效果也显著更差。检测工具只对经批准的研究人员和专家机构开放。
原文 · The Rundown AI
OpenAI said it will embed an invisible watermark in ChatGPT and Codex text across the EU within weeks, under the EU AI Act. "In our testing, watermarking hasn’t affected how capable or fast our models are, or how their responses read." OpenAI also said swapping out words with synonyms can significantly reduce detection, and detection rates were significantly lower for shorter text. Detector tools will only be available to "approved researchers and expert organizations". 💬 4 🔄 0 ❤️ 9 👀 2177 📊 5 ⚡
