01 Quantum has now secured a U.S. patent for technology built to close exactly that gap: a system that lets AI process data while the data stays encrypted from start to finish. The United States Patent and Trademark Office has issued U.S. Patent No. 12,744,764 B1, titled “Secure AI Operations,” covering the core of the Company’s secure AI solutions.

The Confidentiality Problem at the Heart of Enterprise AI

Modern AI is extraordinarily useful and structurally exposed. To get a prediction, a classification, or an analysis out of a model, an organisation typically has to send its raw data into that model — customer records, financial information, proprietary research, operational telemetry. The moment that data is submitted, it exists in a readable form on infrastructure the data owner may not own, staffed by people they’ll never meet, governed by assumptions they can’t verify.

For most workloads that’s an accepted risk. For data-sensitive environments — regulated industries, government, defence — it’s frequently a dealbreaker. The value of AI is obvious; the cost of exposing the underlying data is unacceptable. That tension has quietly kept AI out of many of the environments where it could do the most good.

Computation-Preserving Cryptography: Using Data Without Revealing It

01 Quantum’s patented technology addresses the problem through what the Company calls computation-preserving cryptography (CPC). The principle is deceptively simple to state: sensitive data is encrypted before it is ever submitted to an AI model, it is processed while it remains encrypted, and the result is returned to the user as an encrypted prediction. At no point in that chain is the data exposed in the clear.

This reframes the trust question entirely. The organisation no longer has to trust the AI platform, its operators, or the environment it runs in — because none of them ever see the underlying information. The data owner keeps control of what matters most, even while someone else’s system does the computing. It’s the difference between locking your valuables in a safe you carry with you and simply hoping the hotel is honest.

Protecting the Model, Not Just the Data

The patent covers more than one side of the transaction. Conventional thinking about AI security focuses on protecting the user’s data, but AI models are themselves valuable, proprietary assets — the product of enormous investment, and intellectual property in their own right. 01 Quantum’s technology is designed to protect both: sensitive user data and the proprietary model, together, throughout the inference process.

That dual protection unlocks a genuinely new architecture. The patent also covers a secure AI model marketplace, where model users and model vendors can interact while cryptographic protection is maintained over both the users’ sensitive data and the vendors’ proprietary model IP. In other words, a vendor can offer a model without exposing how it works, and a user can run their data through it without exposing what the data contains. Both parties transact without either surrendering their crown jewels — a foundation for AI collaboration in environments where that has never been safely possible.

Built for the Environments Where Security Is Paramount

This capability is aimed squarely at the settings where conventional AI deployment raises the hardest questions: enterprise, government, and defence. These are the domains where data sovereignty, confidentiality, and security aren’t features to be weighed against convenience — they are non-negotiable conditions of doing anything at all. Encrypted AI inference makes it possible to bring AI into mission-critical work without the compromise that has historically kept it out, from secure data sharing to anomaly detection on information that can never be exposed in transit or in use.

Part of a Deliberate, Patent-Protected Strategy

The newly issued patent doesn’t stand alone. It complements 01 Quantum’s existing intellectual property portfolio and advances a consistent strategy: developing patent-protected technologies designed to secure data, communications, and AI applications for the post-quantum era. Alongside the Company’s IronCAP™ cryptographic engine and its patent-pending ZKP and crypto-locking technologies, secure AI is becoming an increasingly important component of the Company’s commercialization strategy — working technology with formal protection around it, not a promise on a roadmap.

As Andrew Cheung, CEO of 01 Quantum, has put it, the issuance strengthens the intellectual property foundation supporting the Company’s secure AI strategy. As AI becomes embedded in mission-critical environments, he noted, organisations need to benefit from AI without exposing their most sensitive information or proprietary models — a capability the Company sees as carrying significant potential across enterprise, government, and defence applications where data sovereignty and confidentiality are paramount.

The Shift This Represents

For years the implicit bargain of enterprise AI has been a trade: capability in exchange for exposure. The organisations with the most to gain from AI have often been the ones least able to accept that trade — and so they’ve waited. Technology that keeps data encrypted straight through inference dissolves the bargain. You no longer have to choose between using AI and protecting what you feed it.

The future of AI in sensitive environments won’t belong to whoever builds the most powerful model — it will belong to whoever can put that power to work without ever exposing the data behind it, and that is precisely the future 01 Quantum has now patented.