Scam.ai said it has partnered with Qualcomm and launched Halo, an on-device deepfake detection model for live video calls, in an announcement made at Computex 2026 in Taipei. According to TechHQ, Scam.ai was featured at Qualcomm’s booth as part of the event’s Agentic AI track, with Halo positioned as background software that can flag synthetic or AI-generated video in real time on desktop systems.
The immediate significance is not just another security startup launch. For enterprise buyers, Halo points to a more specific category emerging inside Enterprise AI: communication authenticity assurance for live meetings, interviews, and executive calls. If the product performs as described, it would push deepfake defense closer to the endpoint rather than relying on cloud-side inspection or post-incident review.
Scam.ai and Qualcomm Target the Endpoint
TechHQ reported that the Qualcomm partnership gives Scam.ai access to device ecosystem resources and optimization support, enabling Halo to run locally on personal computers without cloud infrastructure. That architecture matters. Running detection on-device can reduce the amount of sensitive meeting video that must leave the endpoint, while also helping with response time during live sessions.
For Qualcomm, the partnership also adds a new enterprise AI workload to the PC stack: real-time trust verification. For Scam.ai, access to Qualcomm’s optimization support could materially shorten the path from prototype to deployable desktop software. That is a notable advantage for a startup operating in a category where inference speed, device compatibility, and user experience are all tied closely to endpoint performance.
Halo’s Initial Use Cases: Recruiting and High-Stakes Calls
TechHQ identified two early user groups: HR and recruiting teams conducting video interviews, and high-value executives including CEOs, CFOs, and venture capitalists who take frequent high-stakes calls. That framing is narrow, but commercially sensible. These are communication moments where a synthetic participant could cause direct operational or financial harm.
The recruiting angle is especially notable. TechHQ reported that identity fraud in video interviews is a documented enforcement risk, and cited figures that only 31% of HR leaders feel equipped to detect it and that deepfake fraud attempts have risen more than 2,000% over the past three years. Those numbers, if borne out more broadly, help explain why a point solution focused on live interviews may find quicker budget justification than a broader platform pitch.
The executive communications use case is different but equally consequential: investor calls, sensitive finance discussions, and external relationship management. In these settings, the cost of a missed synthetic impersonation could be high, but so could the cost of a false alarm.
Why This Matters to Technology decision-makers
For CIOs, CISOs, and collaboration platform owners, the launch is a reminder that deepfake risk is moving from abstract governance debate into endpoint operations. Buyers evaluating tools like Halo should focus less on headline claims and more on deployment realities:
1. Architecture and privacy
An on-device approach may be easier to evaluate in regulated or privacy-sensitive environments because it avoids routing live meeting content through additional cloud infrastructure. That could become a differentiator as organizations tighten controls around AI data handling.
2. Hardware fleet complexity
TechHQ said Halo is optimized for Qualcomm-powered devices. That raises practical questions for enterprises with mixed endpoint estates. Performance, battery impact, manageability, and support coverage may differ across PC classes, making hardware qualification a likely first step rather than broad rollout.
3. Workflow design
Real-time detection introduces a new operational problem: what happens after a flag appears? HR, legal, security, and executive support teams may need new playbooks covering investigation, escalation, participant notification, and call termination criteria.
4. Enterprise readiness
TechHQ reported that more enterprise integration details and additional platform partnerships would be announced in the coming months. That suggests large organizations should expect an evaluation-stage product motion, with open questions around admin controls, telemetry, procurement maturity, and integration into broader security operations.
These are the same kinds of release-rhythm and validation issues now shaping other enterprise AI categories, a pattern also visible in our coverage of Google’s faster enterprise release rhythm and the new enterprise test around AI security responses.
Market Signal: A New Budget Line May Be Forming
Halo’s launch also has implications beyond Scam.ai itself. If enterprises start treating live-call authenticity as a distinct control layer, several adjacent markets could feel pressure. Cloud-centric fraud detection vendors may face new competition from on-device approaches. Unified communications platforms may be pushed to support native authenticity features or deeper integration with third-party detectors. Identity verification providers focused on document checks and onboarding may find that live-session verification becomes a separate buying center.
At the same time, the category remains early. Within the supplied materials, the partnership and product details are effectively single-sourced to TechHQ. A separate TechHQ roundup of generative AI security platforms did not mention Scam.ai, Halo, or the Qualcomm partnership. That does not negate the announcement, but it does mean technology decision-makers should treat the company as an emerging entrant rather than an independently validated market leader.
From a category perspective, Halo sits at the intersection of Models, endpoint security, and collaboration infrastructure. It is a focused bet that enterprises will pay for preventive authenticity checks in narrow, high-consequence moments before they commit to broader anti-deepfake platforms.
Sources and Methodology
This article used a multi-source input set, but the core news event was effectively single-sourced within the provided materials. Specific factual claims about the Qualcomm partnership, Halo, Computex 2026, product behavior, target users, and availability are attributed to TechHQ’s June 29, 2026 report. We also reviewed a separate TechHQ article on generative AI security platforms, which did not mention Scam.ai or Halo; that omission informed our caution on market validation and corroboration.
