DuckDuckGoose AI Review: Deepfake Detection Intelligence Brief for SMBs

Many of the deepfake detection vendors attracting attention today are based in the United States. DuckDuckGoose AI isn't. It was built in Delft, Netherlands, with EU privacy and data sovereignty requirements baked in from day one. Despite being a relatively young company, it has already earned customers in banking, forensic investigation, and government environments where trust and verification are mission-critical.

Disclaimer: All information in this article was verified as of May 2026. Vendors update features, pricing, and benchmarks frequently. These insights reflect the current state of the application. As deepfake and impersonation technologies continue to improve, many of today's limitations, risks, and solutions are likely to change significantly over the next few years.

DuckDuckGoose is a Dutch deepfake detection company founded in March 2020 by three TU Delft students, Parya Lotfi (CEO), Joris Mollinga, and Mark Evenblij. It started as a student project analysing how manipulated media could be used as fake evidence in court, and grew into a company once the police and the Dutch government said yes, they'd actually pay for this.

The company bootstrapped for roughly four years before raising a €1.3M pre-seed in June 2024 from Shaping Impact Group, Ctrl+Alt+Invest, Aruma Ventures, the Graduate Entrepreneur Fund. By 2026, they have raised $2.6 million in funding as mentiond on forbes.

That funding level matters for the SMB conversation: this is a small, focused European team, not a US enterprise giant. Strengths and trade-offs follow from that.

Customers of DuckDuckGoose AI:

The angle DuckDuckGoose leans on, repeatedly, is explainable AI. They don't just return a probability score, they return a visual trace showing which region of an image or which segment of a clip triggered the flag. For a compliance team that has to justify a rejected onboarding to a regulator, that matters.

DuckDuckGoose runs three detection engines, each handling one media type. Together they cover the surface area an SMB actually sees in fraud: selfies, documents, video calls, and voice.

The vendor publishes these numbers across product pages. Treat them as vendor-reported accuracy that might differ on your own dataset.

The technical approach combines three things:

Available on AWS Marketplace as well, which simplifies procurement if your org already buys through AWS.

Three products, each pointed at a different buyer.

DeepDetector: Their flagship product. Real-time image and video deepfake detection, available via API/SDK or on-premise. This is what bunq integrated into its KYC onboarding pipeline. Best fit if you have engineering resource and want detection wired into an existing flow.

Phocus: A web platform sitting on top of DeepDetector. No code required. Fraud analysts, compliance reviewers, and legal teams upload media (image, video, or audio), get a verdict with explainability, and export an audit-ready report. Best fit if your team needs to manually investigate flagged cases or do ad-hoc verification, and you don't want to build a UI yourself.

Waver: audio deepfake detection across 16+ languages. Voice-cloned executive fraud, synthetic speech in payment authorisation calls, contact-centre account-takeover attempts. Real-time, language-agnostic, integrates via API.

Every commercial engagement runs through their sales team. The AWS Marketplace listing exists, but the visible numbers there are placeholder values, not real pricing.

What this means for an SMB evaluation?: you can't kick the tyres on a credit card the way you can with Reality Defender's free API tier. You'll need to book a demo, have a budget conversation, and likely negotiate a pilot. That's a higher-friction entry for organisations doing low-volume evaluation, but it's not unusual for the European B2B SaaS playbook, and it leaves room for negotiating pilot terms.

You may also contact DuckDuckGoose's sales representative, Enrique, at [email protected] for further assistance.

Use these to decide whether DuckDuckGoose belongs on your shortlist.

DuckDuckGoose is a credible, focused European deepfake detection vendor with a real customer base , including a top-tier neobank, a national forensic institute, and a national parliament. Their explainable AI output is the genuine standout: most competitors give you a probability score, DuckDuckGoose gives you a probability score plus a defensible visual artifact, and that gap matters once you're in front of regulators or in court.

For an SMB sitting in the EU and doing digital KYC, identity verification, or any onboarding flow where a synthetic face could pass through, DuckDuckGoose belongs on the shortlist. The GDPR-first posture, on-prem option, and EU footprint are real procurement advantages over US-based alternatives.

The trade-offs are honest ones. There's no self-serve free tier, so you can't pilot it on a credit card the way you can with Reality Defender. The company is smaller than its US-based competitors, with €1.3M raised against Reality Defender's $52M+.

If you're running a US-only stack with a heavy live-call/Zoom requirement and need to procure quickly via a free tier, Reality Defender is the easier entry. If you're EU-based, care about explainability, and want a vendor whose model has been validated inside actual forensic and banking deployments, DuckDuckGoose is worth the demo.

For a demo, simply click the "Get Started" option → fill up the form and submit the request. Alternatively, you can contact DuckDuckGoose's sales representative, Enrique at [email protected] for additional information.