Reken’s On‑Device AI Security Platform Usher in a New Era Against AI‑Powered Phishing and Fraud

Reken, led by former Google ‘click fraud czar’ Shuman Ghosemajumder, is launching an on‑device AI security platform that bypasses cloud‑based analysis in favor of proprietary models running directly on user hardware.

Explaining the On‑Device AI Architecture and Hardware Requirements

The Reken Private Core consists of compact, quantized AI models that can operate on ordinary corporate laptops without GPUs while still delivering real‑time threat detection. Ghosemajumder emphasizes that the team achieved both low resource consumption and full‑speed performance, a combination he claims competitors have not matched.

Northstar Replaces Traditional Security Training

Northstar, the first product built on the Private Core, aims to supplant ineffective security‑awareness training by verifying the legitimacy of message senders rather than merely flagging AI‑generated text.

■The FBI’s Internet Crime Complaint Center recorded $20.9 billion in cyber‑crime losses for 2025, a 26 % year‑over‑year increase.

■AI‑related complaints surpassed 22,000, marking the first time this category was added.

■A 2026 RBC poll shows 83 % of respondents now assume any online message is a scam unless proven otherwise.

Funding and Strategic Partnerships

Founded in 2024 with former Shape Security colleague Rich Griffiths, Reken secured a $10 million seed round led by Greycroft and FPV Ventures.

Network Effects and Privacy Balance

The envisioned Reken Network would link adopters into a “protected circle” where internal communications become mutually verifiable, while threat intelligence is aggregated and anonymized to improve the system without exposing individual data. Ghosemajumder likens this to Apple’s iMessage trusted bubbles but stresses stronger privacy safeguards.

Ghosemajumder warns that AI‑driven fraud has outpaced human perception, making real‑time, on‑device AI the only viable defense. He also argues that conventional security‑awareness programs are futile, forcing employees into impractical forensic roles and wasting time and resources.