Reken's On-Device AI: A New Front in the War on Digital Deception?

SAN FRANCISCO, CA – July 13, 2026 – In an increasingly treacherous digital world where trust is a depreciating asset, a new company is making a bold claim: it can build an internet safe for humans. Today, Reken, an AI cybersecurity firm founded by veterans of Google and Shape Security, emerged from a two-year stealth period armed with $10 million in funding and a novel architecture designed to fight AI with AI, directly on your device.

The timing could not be more critical. The FBI’s latest Internet Crime Report paints a grim picture, with losses surging to a record $20.9 billion in 2025. For the first time, the agency created a specific category for AI-related crime, logging over 22,000 complaints. This statistical reality is mirrored in public sentiment; a recent RBC poll found that a staggering 83% of people now assume any online message is a scam unless proven otherwise. It’s this profound erosion of trust that Reken aims to reverse.

"The Internet is not safe, and is getting less safe every day because of AI," said Shuman Ghosemajumder, Reken's CEO, who previously founded Google's Trust & Safety group and was Head of AI at F5. "Traditional cybersecurity has failed to solve these problems, and increasing scams, fraud, and cybercriminal use of AI have accelerated this erosion of trust. We need a new architecture to prevent our critical online channels from becoming overwhelmed."

The New AI Arms Race on Your Device

Reken's proposed architecture is a fundamental departure from cloud-centric security models. The company's core technology, the Reken Private Core, is an on-device AI security system. Instead of sending data to the cloud for analysis, it uses proprietary AI models to analyze communications and detect threats directly on a user's computer or smartphone. This approach promises two significant benefits: enhanced privacy, as sensitive communication data never leaves the device, and real-time detection, which is crucial for thwarting fast-moving social engineering attacks.

The technical claim is ambitious: the platform runs on commodity hardware without needing specialized GPUs or incurring the variable token costs associated with many large AI models. This efficiency is key to its scalability. By embedding next-generation telemetry sensors, the Private Core is designed to detect not just malicious content but also the subtle fingerprints of AI bots and automation used in sophisticated attacks, from deepfakes to hyper-personalized phishing emails.

This strategy places Reken at the forefront of an emerging trend. As the industry enters the “AI PC” era, with new devices featuring dedicated Neural Processing Units (NPUs), the potential for powerful, on-device security is growing. While established endpoint security leaders like CrowdStrike and SentinelOne already leverage AI, Reken's intense focus on the communication layer and its privacy-first, on-device model represents a specialized and potentially disruptive evolution. The platform creates what Reken calls the 'Reken Network'—a self-assembling trust layer where organizations using the system can communicate with greater confidence, while communications from outside the network are subjected to rigorous analysis.

Beyond Human Error: A New Strategy for the Weakest Link

For years, the cybersecurity industry has preached that humans are the weakest link, responding with an ever-growing regimen of security awareness training. Reken is challenging that premise with its first product, Northstar. Built on the Private Core platform, Northstar is a 'pro-worker' application designed to remove the security burden from employees.

"Companies spend millions of human hours per day and billions of dollars on security training that simply doesn't work," Ghosemajumder stated. "We shouldn't be forcing employees to become forensic digital investigators. We need just-in-time AI that detects the threats the human eye cannot see. That's what Northstar does."

The application works in the background, analyzing incoming communications for signs of social engineering, business email compromise, deepfakes, and other AI-enabled fraud. Instead of asking an employee to spot a phishing link, Northstar aims to identify and flag the threat before the employee even has to make a decision. This shift from user education to automated protection could have profound implications for corporate security strategies and budgets.

The approach has already garnered praise from seasoned industry leaders. Jim Routh, the former CISO of titans like American Express and Aetna/CVS Health, noted the novelty of the solution. "There is nothing like this available to CISOs," he said. "This will enable companies and their supply chains to significantly improve the level of safety they can expect in their communications." This endorsement points to a significant unmet need in the enterprise market, where the cost and ineffectiveness of current anti-phishing measures are persistent pain points.

The 'Google Mafia' Effect: Pedigree, Capital, and a Mission

An ambitious mission requires a formidable team, and Reken’s composition is a significant part of its story. The company is led by Ghosemajumder and Rich Griffiths, who were both instrumental in building Shape Security, the AI bot-defense leader acquired by F5 for $1 billion in 2020. Ghosemajumder's experience dates back to the early days of Google, where he helped launch Gmail and built the company's first Trust & Safety product group, tasked with protecting over a billion users.

This deep-seated expertise attracted a powerful coalition of investors. The $10 million seed round, raised in 2024, was led by Greycroft and FPV Ventures. The investor list reads like a who's who of the 'Google Mafia'—a network of early Google employees and connected investors known for backing transformative tech companies. It includes Wesley Chan of FPV (who founded Google Analytics and Google Voice), Hunter Walk and Satya Patel of Homebrew, and Marcie Vu, the Greycroft partner and former Google IPO leader who now sits on Reken's board.

This concentration of capital and experience is not merely a vote of confidence; it is strategic fuel. The investors bring not just money, but a network and a playbook for scaling complex, AI-driven platforms that operate at a global scale. Their backing suggests a belief that Reken is not just another security vendor, but a company with the potential to build a foundational new layer for internet trust, much as Ghosemajumder’s previous teams did for search and email.

With Northstar now available through an Early Access Program, Reken is transitioning from concept to market reality. The company is betting that its unique blend of on-device AI, a privacy-first philosophy, and an elite founding team is the right formula to finally turn the tide in the escalating war against digital deception.