AI 偵測功能
Google 將 SynthID 擴展至 Chrome Circle to Search
Google 擴展其 SynthID 浮水印偵測系統，讓 Chrome、Circle to Search 及 Google Lens 等工具，得以透過 Google 生成式人工智能系統嵌入的隱形來源元數據，識別出人工智能生成或人工智能編輯的圖像。SynthID 最初為 Google 自家圖像生成模型而推出，其後整合至 Gemini 應用程式之中，會附加隱藏標記，讓兼容的偵測器得以讀取，藉此判斷內容是否於 Google 生態系統內由人工智能工具創作或修改。
此次推行，亦將內容憑證支援，擴展至更多 Google 服務——Gemini 將即時獲得整合功能，而 Chrome 及搜尋則將於未來數月內陸續支援。Google 亦於近期推出的 Pixel 相機系統中，加入原生內容憑證，並表示包括 OpenAI、Kakao 及 ElevenLabs 在內的合作夥伴，計劃為特定人工智能生成內容，採用兼容的 SynthID 標記機制。
此舉反映出，隨著合成媒體內容日益普及且愈趨逼真，業界正加緊努力，將來源追蹤標準化。
英文原文
Google expanded its SynthID watermark-detection system so tools including Chrome, Circle to Search and Google Lens can identify AI-generated or AI-edited images using invisible provenance metadata embedded by Google’s generative AI systems. Originally introduced for Google’s own image-generation models and later integrated into the Gemini app, SynthID appends hidden markers that compatible detectors can read to determine whether content was created or modified with AI tools inside Google’s ecosystem.
The rollout also extends content-credential support across more Google services, with Gemini receiving integrations immediately and Chrome and Search support arriving in the coming months. Google additionally added native content credentials to recent Pixel camera systems and said partners including OpenAI, Kakao and ElevenLabs plan to adopt compatible SynthID tagging for certain AI-generated outputs.
The move reflects growing industry efforts to standardize provenance tracking as synthetic media becomes more widespread and convincing.
- 來源
- Trend Hunter
- 發布
- 2026-05-25
- 品類
- Internet
- 出處
- engadget, deepmind.google