Integrated Quantum Technologies Inc. (IQT) is an enterprise AI infrastructure company developing technologies designed to enable organizations to use sensitive data with artificial intelligence while reducing exposure of the underlying information. The Company’s primary commercial product, VEIL™ (Vector-Encoded Information Layer), is a privacy-preserving technology designed to protect sensitive information across AI and machine learning environments.
Enterprise AI has created an uncomfortable imbalance, universally. Regardless of industry, organizations are moving quickly to put more data to work, while parallel pathing ways to keep it secure.
The benefits of AI are clear. Actualizing them, on the other hand, presents significant challenges. This can require the introduction of new technologies, connecting them with existing systems and data, determining where and how they should be used, establishing appropriate governance, and ensuring they are effectively adopted across the organization. All of this requires time, resources and coordination across multiple parts of a business.
Security and privacy add in complexity — and enterprises are still learning how to navigate. Many enterprise safeguards for data were developed before today’s generation of AI. They remain important, but organizations do themselves a disservice if they assume these safeguards will account for every risk, present and future, that AI introduces.
That means building on what already works while also rethinking some of the assumptions that have informed today’s approaches.
To start, does AI need access to sensitive data in its original form at all?
According to Check Point’s 2026 report, organizations globally experienced an average of nearly 2,000 cyberattacks per week in 2025. Add to that McKinsey’s 2025 global survey, which found that regular AI use in at least one business function was at 88% in organizations, up from 78% a year earlier. Use of AI is growing, so is the amount of data being accessed, shared and processed, increasing the potential exposure and risk.
Our instinct is to find better ways to protect that data. Encryption, differential privacy, federated learning and other privacy-enhancing technologies can reduce risk as sensitive data moves through the AI lifecycle. But protection is only one important consideration.
Stopping there assumes that exposure is unavoidable. But what if it isn’t?
Instead of asking how to protect sensitive data after it enters the AI ecosystem, we should take two steps back and also ask whether the original data needs to enter that environment at all.
Exposure does not have to be the starting point
Before sensitive data enters an AI workflow, a number of questions should be part of organizational due diligence.
- What information does the model need to perform the task?
- Which sensitive or identifying details, if any, are necessary?
- Does that information need to be provided in its original form?
The goal should be to determine what information the task actually requires — and what it doesn’t.
If an AI system can perform its intended task using only the information it needs, there is an opportunity to reduce what is exposed before that information moves further into the AI ecosystem.
This changes the conversation. Instead of assuming sensitive data needs to be available and then determining how best to protect it, reducing exposure becomes part of the architecture itself.
This is the approach behind VEIL, developed by Integrated Quantum Technologies (IQT). VEIL transforms sensitive data into non-invertible representations before it enters downstream AI workflows, allowing AI systems to use the information required for an approved task without exposing the original sensitive data.
Industry and enterprise both have a role to play
Reducing exposure by design cannot rest solely with the organizations adopting AI. Industry has a responsibility to develop security and privacy solutions designed for the realities of AI. Enterprises, meanwhile, need to consider these issues from the outset, rather than addressing them once new technologies and workflows are already in place.
We are asking AI to do things that weren’t possible even a few years ago. That same ambition should extend to how we think about how it is secured. Protecting what is exposed will always matter, but so will reducing unnecessary exposure in the first place.
AI gives us an opportunity to reconsider some of the assumptions that have shaped data security for years. We should take it. Because when we stop questioning those assumptions, we create vulnerabilities and give adversaries the advantage.
Visit https://www.integratedquantum.com/ for more information.
Related News:
Q&A: Anita Oehley of Integrated Quantum Technologies on Scaling Secure AI
