Press Release · Validix
Validix Unveils Groundbreaking Technology to Make AI-Generated Content Verifiable, Reliable and Accountable
Distributed by AdWebGo · Originally issued via PR Newswire
New patent-pending innovation gives regulated industries traceable and verified AI output.
ZUG, Switzerland, Nov. 17, 2026 — Validix, a company advancing governed claims for enterprise AI, today unveiled its Atomic Claim Architecture, a new patent-pending invention that changes how organizations govern content produced by generative AI.
The Validix Atomic Claim Architecture is a secure governance layer, independent of and complementary to large language models (LLMs). It constrains what they can generate and verifies content before it is released. The model itself is not modified, so organizations can apply the architecture to AI tools they already use or plan to deploy in the future.
Generative AI writes quickly and convincingly, but it can also get facts wrong or invent them in the case of hallucinations. In regulated industries, a single unsupported statement can lead to regulatory action, financial loss, legal exposure and reputational damage, and the risk grows with every piece of content AI produces.
Validix provides continuous monitoring of the evidence. At a practical level, this means that marketing and sales teams can verify communications through the Validix platform to check that the information is supported by current evidence.
Why this matters for regulated industries
Across regulated industries, the accuracy and reliability of information is not just a matter of trust, but a critical business and regulatory requirement. In the pharmaceutical sector alone, companies have incurred more than $66 billion* in recorded regulatory penalties and settlements globally since 2010, highlighting the significant financial consequences of compliance failures, including misleading claims, inaccurate information or non-disclosures.
Lluis Pallarès, co-founder of Validix, said: "Every organization using AI to communicate is putting its name behind what the technology says. That responsibility cannot be delegated to a language model. Our patent-pending architecture turns evidence into enforceable boundaries for what AI is allowed to assert, and makes release a separate, controlled decision, without requiring changes to existing AI models. AI generates. Validix governs."
"Validix is designed to accelerate expert review, not replace it: reviewers see the evidence behind every claim and keep the final say," adds Pallarès.
Validix's first application, built for teams that create, review and monitor content in life sciences, is already in live use at a pharmaceutical company following a pilot project.
Validix plans to launch its commercial platform in early 2027 and is opening a limited number of pilot partnerships across regulated industries.
How does it work?
The Validix architecture breaks evidence into "Atoms," which are small, verifiable units of truth, each traceable to its source. Every claim is built from these Atoms, so it can be traced to the evidence behind it. If AI-generated content cannot meet this standard, it is withheld or sent for human review.
Validix also monitors the evidence continuously. If the underlying source evidence changes, every affected claim is flagged and sent to a human reviewer, so organizations stay in control and published statements stay aligned with the current evidence. Together, upstream control, pre-release checks and continuous monitoring of evidence sources form what Validix calls a "Governed Claims Infrastructure," a new category of infrastructure dedicated to governing claims made with AI.
*Violation Tracker Global: Current Parent Industry Penalty Totals, Source: Violation Tracker, The website is produced by Good Jobs First, a US-based nonprofit organization focused on corporate accountability and government transparency. United States False Claims Act settlements and judgments exceeded $6.8B last year, Source: DOJ
SOURCE Validix
About Validix
Validix AG is a Swiss technology company based in Zug, building Governed Claims Infrastructure for enterprise AI. Its patent-pending Atomic Claim Architecture connects supporting evidence, individually governed claims, generation constraints and independent output validation to help organizations control what AI is permitted to assert on their behalf.
Video explainer on how Validix Atomic Claim Architecture works: https://validix.ch/how-validix-works.
For more information visit our website (validix.ch) or follow us on LinkedIn
To request a demonstration https://validix.ch/contact