Real-time deepfake and AI-generated content detection for secure digital interactions.
Free plan available, paid tiers aboveFreemiumSecurity
Visit Website FreemiumVisit Website
Tracked since2026
The Bottom Line
- Price
- Free plan available, paid tiers above. Compare plans
Key facts
- Detects deepfakes and AI-generated content across images, video, and audio.
- Offers real-time, on-device deepfake detection for live video calls (Halo).
- Provides a single API for both manipulated reality and fully synthetic content detection with high accuracy.
Pros
- Detects both manipulated deepfakes and fully synthetic AI content with one platform.
- High detection accuracy (98.2%) and fast processing times (under 4 seconds).
- On-device detection for live calls ensures privacy and real-time protection.
- Flexible, usage-based pricing with a free tier and no contracts.
- Easy integration with a comprehensive API and no-code options.
Cons
- Voice detection for live calls is on the roadmap, not fully implemented for Halo.
- Forensic PDF reports are significantly more expensive ($500 per image).
- The free tier is limited to 200 images per month with the faster, less forensic model.
What is Halo by Scam AI?
Scam AI offers advanced deepfake and AI-generated content detection solutions, primarily through its Eva models and the Halo on-device detection tool. It is designed to identify both manipulated real people (deepfakes like face swaps, lip-sync, voice clones) and fully synthetic content generated by AI models such as Stable Diffusion, DALL·E, Midjourney, and Sora, across images, video, and audio. The platform provides real-time detection with high accuracy (98.2% for visual detection in under 4 seconds) and is built for various industries including financial services, contact centers, hiring, media, and government. Halo, specifically, offers on-device deepfake detection during live video calls (Zoom, Teams, Meet) to flag synthetic faces and faceswaps without recording or sending data. Scam AI aims to combat AI-generated fraud, projected to cost businesses significantly, by providing a comprehensive, easy-to-integrate API solution with transparent, usage-based pricing.
Key Features
- Deepfake detection (face swaps, lip-sync, reenactment, voice clones)
- Generative AI content detection (images, video, audio from DALL·E, Midjourney, Stable Diffusion, etc.)
- Real-time detection during live calls (Halo on-device)
- Visual detection for images and video (98.2% accuracy)
- Voice and audio clone detection
- Identity and document verification (KYC, IDV)
- RESTful API for integration
- Detailed analysis reports with confidence scores and heatmaps
Pricing Plans
Pricing checked Oct 1, 2026
| Plan | Price | Details |
|---|---|---|
| Self-Serve | $0 / image (200 free images/month) |
+4 moreShow less
|
| Enterprise | Contact us |
+4 moreShow less
|
| Forensic PDF Reports | $500.00 / image |
+2 moreShow less
|
Is Halo by Scam AI worth the price?
Good value
The Self-Serve tier is very fair, offering 200 free images monthly and then a reasonable $0.05/image.
This makes it highly accessible for individual users and small businesses. The $500 per image for Forensic PDF Reports is expensive but justifiable for legal or high-stakes investigations.
This structure is best for a wide range of users, from casual to enterprise, with specific needs for deepfake detection and forensic analysis.
Hidden Costs & Gotchas
Overage fees apply after 200 free images/month.
Optional add-ons increase per-image cost.
Reviews
$99Free with your review
Write a reviewReview Halo by Scam AI, get a free AI guide
Share your experience and we will send you Improve Your Thinking Patterns Using ChatGPT, free.
Explore More
Halo by Scam AI FAQ
How does Halo's on-device deepfake detection work during live calls?
Halo runs 100% on your device, flagging synthetic faces and faceswaps in real-time during live video calls on platforms like Zoom, Teams, and Meet. It does this without recording or sending any data off your device, ensuring privacy.
What is the difference between Eva-v1-Fast and Eva-v1-Pro models?
Eva-v1-Fast processes images in under 2 seconds and is included in the free tier, suitable for real-time moderation. Eva-v1-Pro delivers forensic-grade accuracy in under 4 seconds, designed for more critical applications like KYC and legal evidence, and is available in the Enterprise plan.
Can Scam AI detect deepfakes in audio, and is this available for live calls?
Yes, Scam AI can detect voice clones and synthetic speech through its API for audio files. For live calls, voice detection is currently on the roadmap for the Halo on-device product, with video detection being the primary focus for live calls at present.
What types of AI-generated content can Scam AI identify beyond deepfakes?
Beyond deepfakes (manipulated real people), Scam AI can detect fully synthetic content generated from scratch by modern AI models such as Stable Diffusion, DALL·E, Midjourney, Sora, and ElevenLabs, across images, video, and audio.
How does Scam AI ensure data privacy and compliance?
Scam AI is GDPR compliant and SOC 2 Type II attested. By default, it does not retain media after processing. Enterprise customers have the option to configure custom retention settings for audit and compliance purposes. Halo's on-device detection specifically processes data locally without sending it to the cloud.
What are the optional add-on features and what do they provide?
Optional add-on features include Adaptive Defense for real-time GenAI, deepfake, and injection attack detection; Active Liveness for verifying real human presence; and Express Lane for low-latency processing with a 3-second response time guarantee. These enhance detection capabilities for specific use cases.
Source: scam.ai