Halo

What is Halo?

Halo is a real-time deepfake detection tool from Scam.ai that runs entirely on your Windows PC. It analyzes every face on live video calls—Zoom, Microsoft Teams, Google Meet, WebEx, and Slack—and flags synthetic faces or face-swaps the moment they appear. Finance and HR teams use it to verify they're talking to the real person before approving wire transfers or making hiring decisions. The entire detection process happens on-device: no recording, no uploads, no third-party servers.

Application scenarios

Wire transfer approval
Flags synthetic faces on live calls before a finance employee authorizes a transfer, targeting fraud like the Arup case where a deepfake CFO led to ~$25M in losses.
Executive assistant screening
Protects assistants and execs who routinely join calls with leadership and may not recognize a subtle face-swap.
HR interviews
Detects face-swaps during remote interviews to prevent bad hires from candidates using synthetic video.
Treasury and accounts payable
Monitors routine approval calls where fraudsters impersonate CFOs or colleagues to greenlight payments.
Compliance and audit
Leaves an on-device audit trail for fraud investigations and regulatory review.
Slack and WebEx calls
Extends protection beyond the big three conferencing platforms to internal and external chat-based video calls.

Core Features

Real-time on-device detection
Halo scans every face in the video frame-by-frame, roughly four times per second, and delivers a verdict while the call is still happening.
No cloud round-trip
Detection happens locally in the same moment as the call—no upload, queue, or download cycle, so alerts land before money moves.
Private by design
Video frames are analyzed in memory and discarded; nothing is recorded, stored, or transmitted to external servers.
Minimal attack surface
No network calls during detection, no frames in transit, and no third-party storage—fewer moving parts than cloud-based detectors.
System tray integration
Installs once on a Windows PC and sits in the tray next to your conferencing app—no plugins or browser extensions required.
Cross-platform call coverage
Works across Zoom, Microsoft Teams, Google Meet, WebEx, and Slack calls.
Synthetic media detection
Flags synthetic faces and face-swaps specifically, not just liveness checks—it protects the whole call, not just the login moment.
On-device audit trail
Keeps a local record of detections for fraud and compliance review.

Target users

Halo is built for finance and treasury teams—including accounts payable, exec assistants, and treasury staff—who approve money movement on video calls. HR and recruiting teams also benefit by verifying candidate identities during remote interviews. The tool is designed for organizations where a single deepfake call could cost millions or compromise hiring integrity.

How to use Halo?

Download Halo from the official Scam.ai site and install it on your Windows PC. It runs in the system tray alongside your conferencing app—no plugins or browser extensions. Once installed, it automatically scans every face on your live calls. If a synthetic face is detected, an alert surfaces during the call, giving you time to stop and verify the person's identity. For detailed setup instructions, visit the official website at https://www.scam.ai/en/halo.

Effect review

Halo addresses a genuine and growing threat: real-time face-swaps that bypass traditional security controls like email filters and login checks. The on-device architecture is a strong design choice—it eliminates the latency and privacy concerns of cloud-based detection, which matters for sensitive financial calls. The Arup case cited on the site (a ~$25M loss from a deepfake CFO call) gives the tool concrete real-world urgency. The frame-by-frame scanning at four times per second suggests detection is fast enough to be actionable mid-call. For finance and HR teams that already worry about social engineering, Halo offers a practical, privacy-preserving layer of defense. The lack of user testimonials or independent benchmarks on the site means you should test it in your own environment, but the feature set is well-aligned with the stated threat model.

Frequently Asked Questions

How does Halo detect deepfakes on live calls?

Halo uses on-device AI to analyze video frames in real time, looking for subtle artifacts and inconsistencies in facial features that indicate synthetic manipulation, without sending data to the cloud.

Which video conferencing platforms does Halo support?

Halo integrates with Zoom, Microsoft Teams, and Google Meet, providing real-time deepfake detection directly within these platforms.

Is Halo suitable for finance and HR teams?

Yes, Halo is built for finance and HR teams to verify identities during remote interviews, client calls, or sensitive transactions, reducing the risk of fraud and social engineering attacks.

Does Halo require any special hardware or software?

No, Halo runs entirely on your existing device (laptop or desktop) with a standard webcam. It uses local processing, so no additional hardware or cloud services are required.

Can Halo work during poor internet connections?

Yes, since Halo processes video locally on your device, it remains effective even with low bandwidth or intermittent internet, as it does not rely on cloud analysis.

Is Halo privacy-friendly?

Absolutely. Halo processes all video data on-device, meaning no call recordings or face images are stored or transmitted. It ensures compliance with data protection regulations like GDPR and HIPAA.