Google's SynthID-Text, the first ever production-ready generative watermark system for large language model, designs a novel Tournament-based method that achieves the state-of-the-art detectability for identifying AI-generated texts. The system's innovation lies in: 1) a new Tournament sampling algorithm for watermarking embedding, 2) a detection strategy based on the introduced score function (e.g., Bayesian or mean score), and 3) a unified design that supports both distortionary and non-distortionary watermarking methods. This paper presents the first theoretical analysis of SynthID-Text, with a focus on its detection performance and watermark robustness, complemented by empirical validation. For example, we prove that the mean score is inherently vulnerable to increased tournament layers, and design a layer inflation attack to break SynthID-Text. We also prove the Bayesian score offers improved watermark robustness w.r.t. layers and further establish that the optimal Bernoulli distribution for watermark detection is achieved when the parameter is set to 0.5. Together, these theoretical and empirical insights not only deepen our understanding of SynthID-Text, but also open new av
We introduce SynthID-Image, a deep learning-based system for invisibly watermarking AI-generated imagery. This paper documents the technical desiderata, threat models, and practical challenges of deploying such a system at internet scale, addressing key requirements of effectiveness, fidelity, robustness, and security. SynthID-Image has been used to watermark over ten billion images and video frames across Google's services and its corresponding verification service is available to trusted testers. For completeness, we present an experimental evaluation of an external model variant, SynthID-O, which is available through partnerships. We benchmark SynthID-O against other post-hoc watermarking methods from the literature, demonstrating state-of-the-art performance in both visual quality and robustness to common image perturbations. While this work centers on visual media, the conclusions on deployment, constraints, and threat modeling generalize to other modalities, including audio. This paper provides a comprehensive documentation for the large-scale deployment of deep learning-based media provenance systems.
Recent advances in LLM watermarking methods such as SynthID-Text by Google DeepMind offer promising solutions for tracing the provenance of AI-generated text. However, our robustness assessment reveals that SynthID-Text is vulnerable to meaning-preserving attacks, such as paraphrasing, copy-paste modifications, and back-translation, which can significantly degrade watermark detectability. To address these limitations, we propose SynGuard, a hybrid framework that combines the semantic alignment strength of Semantic Information Retrieval (SIR) with the probabilistic watermarking mechanism of SynthID-Text. Our approach jointly embeds watermarks at both lexical and semantic levels, enabling robust provenance tracking while preserving the original meaning. Experimental results across multiple attack scenarios show that SynGuard improves watermark recovery by an average of 11.1\% in F1 score compared to SynthID-Text. These findings demonstrate the effectiveness of semantic-aware watermarking in resisting real-world tampering. All code, datasets, and evaluation scripts are publicly available at: https://github.com/githshine/SynGuard.
Governments are increasingly mandating that LLM-generated content carry watermarks. The EU AI Act calls for markings that are "sufficiently reliable and robust." California's SB 942 requires disclosure that is "permanent or extraordinarily difficult to remove." Both mandates rest on an untested assumption: that watermark detection yields evidence reliable enough for courts. This paper tests that assumption directly. We evaluate three representative LLM watermarking methods -- KGW, Unigram, and the MarkLLM implementation of SynthID-Text -- against the Daubert admissibility criteria and the NIST SP 800-86 digital forensic process. To structure this evaluation, we propose a Forensic Readiness Score (FRS) framework with 12 criteria, three mandatory gates, and a 60-point scoring system. We focus on meaning-preserving paraphrase as the attack vector, since it is both legally realistic and difficult to dismiss as evidence tampering. The results raise serious evidentiary concerns. Out of 846 valid paraphrase runs across 15 diverse prompts per method, every single initially-detected KGW and Unigram text lost its watermark after paraphrasing -- 100% conditional removal. SynthID fared only sl
We introduce TextSeal, a state-of-the-art watermark for large language models. Building on Gumbel-max sampling, TextSeal introduces dual-key generation to restore output diversity, along with entropy-weighted scoring and multi-region localization for improved detection. It supports serving optimizations such as speculative decoding and multi-token prediction, and does not add any inference overhead. TextSeal strictly dominates baselines like SynthID-text in detection strength and is robust to dilution, maintaining confident localized detection even in heavily mixed human/AI documents. The scheme is theoretically distortion-free, and evaluation across reasoning benchmarks confirms that it preserves downstream performance; while a multilingual human evaluation (6000 A/B comparisons, 5 languages) shows no perceptible quality difference. Beyond its use for provenance detection, TextSeal is also ``radioactive'': its watermark signal transfers through model distillation, enabling detection of unauthorized use.
Artificial Intelligence (AI) models, at their core, apply general learnings from broad datasets to individual circumstances using probabilistic behaviour. This inductive approach stands in contrast to deductive reasoning approaches which seek to prove conclusions from their premises. However, research has shown that deductive reasoning with AI models is a challenging problem and in the real-world it may not always be feasible. An alternative way forward is to leverage abductive reasoning, seeking to corroborate the output of multiple approaches to identify the most likely conclusion from the factual matrix. We apply this to synthetic media detection in forensic settings, and find we are able to disproportionately lower the risk of false positives to true positive recall. We also provide the first empirical evaluation of OpenAI's rollout of SynthID on synthetic images and evaluate how complementary different synthetic media detection approaches are.
Researchers have created self-destructing living plastic that uses engineered bacteria to completely break itself down when activated。 The material degrades in just six days without creating microplastics, offering a potential new solution for single-use plastic waste
Rice University chemists have found a new way to make neodymium, a rare-earth metal, interact with oxygen。 Using a specially designed molecular structure described as a “basket,” the team positioned the atoms so they could form a bond once thought unlikely。 The breakthrough produced highly reactive compounds that could eventually give chemists alte
A quantum problem once described as impossible for classical computers has now been solved using relatively modest hardware。 Researchers used tensor networks to compress the overwhelming wave function created by hundreds of entangled qubits, allowing some calculations to run on a laptop。 Their results matched both theoretical predictions and simula
Scientists say new technologies have reopened the debate over whether Mars could someday be terraformed, turning a once impossible idea into a serious research topic。 Before anyone tries to reshape the Red Planet, though, researchers say we must understand the risks, including what might be lost if Mars already harbors its own forms of life
Four nearby white dwarf stars have been discovered hiding in plain sight beside brighter red dwarf companions。 Hubble's ultraviolet observations finally revealed the long-hidden stellar remnants, including one just 25 light-years away that took nearly three decades to confirm。 The findings match long-standing predictions and suggest our corner of t
MIT engineers have found a way to give chip-based lidar a wider, clearer view without relying on moving parts。 Their design uses differently shaped antennas that can sit close together without scrambling one another’s signals。 In tests, the system sharply reduced interference while steering a single precise beam across a broad field of view
A new AI-powered blood test could give people a remarkably early warning of serious heart and circulation problems。 Developed by researchers at the University of Hong Kong, CardiOmicScore analyzes thousands of proteins and metabolites to estimate the risk of six major cardiovascular diseases, including heart attack, stroke, heart failure, and atria
JWST has captured unusually detailed images of gas feeding the supermassive black hole at the center of NGC 4696。 A vast filament appears to funnel material into an 800-light-year-wide spinning disk, where gas races around at up to 600 kilometers per second。 The findings suggest black holes may recycle their own fuel by heating gas with jets and la
The asteroid that wiped out the dinosaurs was likely an exceptionally rare CO chondrite from a distant region of the solar system。 Its unusual chemistry suggests that planet-cooling dust and debris, rather than sulfur inside the asteroid, may have delivered the deadliest blow
Researchers are applying evolutionary theory to cancer by changing treatments before tumors have time to develop resistance。 Mathematical models suggest that rapid, carefully timed switches between multiple therapies could improve cure rates