synthid

Apple 发布 iOS 27.2 更新，为 iPhone Duo 带来 Apple Pencil 支持、Edge Light 面部照明功能，并扩展 Siri AI 到更多语言，同时在照片应用中支持 Google 的 SynthID 水印检测。

特朗普在 Truth Social 发布炫耀白宫修缮成果的照片，其中一张被 Google 的 SynthID 检测器标记为由 OpenAI 工具生成或编辑，照片中的台阶数量和反射细节均存在明显不一致。

Google DeepMind's SynthID AI watermarking technology has been named one of TIME's Best Inventions of 2026, with over 180 billion images and videos watermarked to help people distinguish AI-generated content from real media.

Google is expanding its SynthID Detector tool globally, allowing anyone to check whether an image, video, or audio file was AI-generated, covering content from Google and partners like OpenAI, NVIDIA, Kakao, and soon Apple.

Google DeepMind 推出 SynthID Detector，可快速识别视频、音频和图片中的水印，即使经过滤镜等编辑处理后依然有效，并可与 Google Chrome、搜索和 Gemini 应用中的现有 AI 识别功能配合使用。

Google has released an improved version of its SynthID AI content detector, now available globally with a login requirement and daily usage quota, supporting detection of watermarks from Google, OpenAI, and Apple.

Google launched a public website that lets anyone verify whether images, videos, or audio were AI-generated using its SynthID watermarking technology, expanding access beyond journalists and researchers. The tool is supported by partners including OpenAI, Nvidia, and Kakao, and handles verification requests at a rate of 1 million per day.

In this episode of the Google DeepMind podcast, the team discusses SynthID AI watermarking technology: from how text and image watermarks work (injecting a bias into token distributions using a secret key) to their applications in biological sequence provenance and biosecurity, as well as the limitations of watermarking for short texts, publicly available detectors, and similar scenarios.

Google DeepMind's podcast explores how imperceptible watermarking like SynthID can help verify whether content was made by humans, machines, or nature, covering media watermarking and SynthID Bio for biological designs.

Google DeepMind published a research paper on a protein watermarking system, adapting SynthID-style probability-biasing watermarks to AI-designed protein sequences so that trusted research proteins can be identified, helping mitigate biosecurity risks from AI protein design tools.

Google DeepMind announced SynthID Bio, a watermarking technology that embeds verifiable signatures into AI-designed proteins while preserving their biological function, verified in wet-lab tests and built into AlphaFold 3's diffusion network.

Apple announced Apple Reference Image for iPhone 18 Pro, a feature that uses signed sensor data and Private Cloud Compute to verify photo authenticity and detect AI alterations.

The article raises concerns about the privacy implications of AI watermarking technologies like SynthID, questioning if tools exist to anonymize or make such content appear human-written.

An author tests the survivability of statistical text watermarks through various transformations, finding that translations and synonym edits often leave the mark intact, but full paraphrase removes it at the cost of fact accuracy.

Google will allow users to remove visible watermarks from AI-generated images, videos, and songs, while maintaining invisible watermarks like SynthID for transparency. The change aims to balance creative control with safety, and includes open-sourcing a new library called Credentio for content credentials.

A gentle, visual walkthrough explaining how statistical watermarking hides a secret mark in AI-generated text by subtly biasing token choices, and how editing can erase it.

The article argues that text AI watermarks are inherently trivial to remove, exploring the EU AI Act's watermarking requirements and the technical challenges of text steganography, including Google's SynthID approach.

Google Earth briefly released an AI feature for creating fake satellite images, then quickly retracted it over concerns it could undermine trust in genuine satellite imagery and enable misinformation. Experts question whether Google's SynthID watermarking is sufficient to prevent abuse.

Google Earth's new AI image generator (Nano Banana) can create realistic altered satellite and aerial images, raising concerns about misinformation despite SynthID watermarks. After Digital Digging demonstrated believable fakes, Google announced it is rolling back the feature.

Google's SynthID watermark is robust against degradation but cannot fully solve AI misinformation; the article tests its durability and discusses the limitations of watermarking and metadata approaches.