F-Secure (Nasdaq Helsinki: FSECURE) and AMD Silo AI, an AI lab that became part of AMD (NASDAQ: AMD) following a 2025 acquisition, announced in a press release on September 8 that they have jointly demonstrated a security approach designed for a specific, increasingly common scenario: AI agents that browse, compare products, manage subscriptions, and complete purchases on a person’s behalf. The approach underpins F-Secure TrustPath, an upcoming product expected to enter beta in the fourth quarter of 2026 and reach production in 2027, meaning what’s being announced here is a technical demonstration and product roadmap, not a shipping consumer product yet, and its real-world effectiveness has not been independently tested.
The Actual Problem This Targets
As AI agents take on more autonomous tasks, reading web content, comparing prices, and eventually completing a purchase or login, they inherit a new kind of vulnerability: the same manipulated reviews, hidden instructions embedded in a web page, or unsafe decision paths that could mislead a person can also mislead an AI agent acting on that person’s behalf. F-Secure’s core design principle here is that trust should be evaluated per moment, not per domain, meaning a website a person generally trusts can still contain untrusted elements, and a routine browsing step becomes meaningfully more sensitive the instant an agent starts acting on the page, such as entering payment details or logging in. That distinction matters because most existing security tools were built around the older model of protecting a human reading a page, not an autonomous system taking action based on what that page contains, which is a meaningfully different threat surface.
How the Routing Approach Actually Works
Rather than granting blanket trust to an entire website, the joint approach evaluates each piece of data individually and decides where it should be processed, on the user’s own device or in the cloud, and which AI model handles it, based on the data’s sensitivity, processing cost, the capability required for that specific task, and performance needs. F-Secure contributes the trust framework spanning a consumer’s full digital journey; AMD Silo AI contributes the compute infrastructure that can dynamically shift processing between on-device and cloud-deployed AI models according to that routing policy. This is a similar underlying trade-off to what shows up elsewhere in AI hardware and privacy design: processing locally on a device tends to be more private and immediate but works with smaller, less capable AI models, while cloud processing unlocks more capable models at the cost of sending data further away and adding latency. The routing approach here is built to make that trade-off dynamically, case by case, rather than committing entirely to one approach or the other, so a low-stakes product comparison might stay entirely on-device while a payment confirmation step gets routed to more capable cloud infrastructure with additional scrutiny, a distinction that requires the system to correctly judge sensitivity in the first place, something the release doesn’t explain the mechanics of in detail.
The Survey Data Behind the Premise
F-Secure cites its own research finding that 84% of consumers worry AI makes it difficult to tell what’s real online, and separately that 44% say they’re worried about using AI at all. A third figure, that only 17% of consumers are currently ready to let AI take actions on their behalf, is attributed to YouGov, an independent polling firm, making it a somewhat more independently sourced data point than the other two figures, which come from F-Secure’s own commissioned research and should be read as the company’s own market framing rather than fully neutral survey data. Taken together, though, all three numbers point in the same direction: consumer comfort with AI acting autonomously online remains genuinely low, which is presumably the market gap F-Secure and AMD Silo AI are positioning TrustPath to address, whether or not the specific percentages hold up to independent scrutiny.
What’s Company Framing vs. What’s Demonstrated
F-Secure CEO Timo Laaksonen and CTO Santeri Kangas both frame this collaboration in significant terms, describing it as taking a stated principle “from principle to practice” and calling it a meaningful step forward for consumer digital security. AMD Silo AI’s Jaakko Vainio framed the underlying philosophy more narrowly: the goal, in his words, is handling each event as close to the user as possible rather than routing everything to the largest available AI model. Those are the companies’ own characterizations of a technology approach that has been demonstrated but not yet shipped in a production consumer product; TrustPath’s actual real-world performance, adoption, and effectiveness against real manipulation attempts remain to be seen once it reaches beta and, eventually, production. The release also notes this collaboration follows a series of other AI-focused partnership announcements from F-Secure, including a prior deepfake-detection partnership with IdentifAI, suggesting a broader pattern of the company positioning itself specifically around AI-era security rather than this being an isolated initiative.
Sources
F-Secure and AMD Silo AI Demonstrate Unique Approach to Helping Secure Digital Journeys in Agentic AI Era, PRNewswire, September 8, 2026.
Editorial Disclosure
This article is based on a press release issued by F-Secure on September 8, 2026, distributed via PRNewswire. Securities discussed: F-Secure Corporation (Nasdaq Helsinki: FSECURE) and Advanced Micro Devices, Inc. (NASDAQ: AMD), of which AMD Silo AI is a part. Next Gen Tech Stocks has not received compensation from F-Secure, AMD, their management, investor relations representatives, or any third party for this coverage. No staff member or principal of Next Gen Tech Stocks holds a position in these securities at the time of publication. Two of the three survey statistics cited originate from F-Secure’s own commissioned research rather than independent third-party polling. F-Secure TrustPath is an upcoming product expected to enter beta in Q4 2026 and production in 2027; statements regarding its timeline and eventual real-world performance are forward-looking and unproven. References to these companies are for market context and analytical purposes only and do not constitute an investment recommendation. All securities carry investment risk including possible loss of capital. Coverage on Next Gen Tech Stocks is for informational and educational purposes only. See our full Disclaimer.
