VIDEO PODCAST
Integrated Quantum Technologies' Jeremy Samuelson: AI Security and Quantum Threats
James Maguire
June 14, 2026

As enterprises race to deploy AI, many are discovering that model performance is only part of the challenge. Data quality, governance, security, and regulatory compliance have become equally important concerns, particularly in industries handling sensitive information.

In this interview, Jeremy Samuelson, Executive Vice President of AI and Innovation at Integrated Quantum Technologies, explains why most AI initiatives ultimately succeed or fail based on data strategy, explores the growing security risks surrounding AI systems, and introduces Veil, the company’s platform designed to protect sensitive information without sacrificing model accuracy.

Samuelson argues that traditional approaches like homomorphic encryption and differential privacy force organizations to choose between security, performance, and cost. Veil takes a different approach by transforming data into compressed, anonymized representations that preserve predictive power while eliminating sensitive information.

He also discusses the emerging threat of quantum-enabled attackers, why organizations should prepare sooner than they think, and how Integrated Quantum Technologies is bringing Veil directly to Snowflake customers through its new marketplace application.

Core Takeaways
AI Success Depends on Data
Despite the excitement surrounding generative AI, organizations continue to face the same challenge that has existed for decades: poor-quality data undermines AI performance, making data governance and trust foundational to every successful AI initiative.
A Growing AI Security Gap
Most organizations separate AI teams from cybersecurity teams, creating a dangerous gap where neither group fully understands the emerging attack surfaces introduced by modern AI systems.
Protecting Data Without Trade-Offs
Veil uses irreversible data encoding and compression to preserve predictive accuracy while removing sensitive information, avoiding the performance penalties of homomorphic encryption and the accuracy compromises of differential privacy.
Preparing for Quantum Threats Today
Quantum-enabled attacks may arrive sooner than many expect, and organizations handling sensitive information should begin adopting security architectures that remain effective even if traditional encryption is eventually broken.
KEY QUOTES

Every AI Initiative Is Really a Data Initiative

There's this big push usually from leadership within a company to implement AI. They're like, 'We got to be doing AI. Let's get to the AI.' The thing is, whatever flavor of AI you want to talk about, all the way back to classical statistical models, classical machine learning, deep learning, they've all had the same problem. All of these systems have always had the problem of garbage in, garbage out.
If you don't have clean, reliable data that people trust, that people actually trust the answers you get, then your whole AI initiative is up in smoke. AI gets the headlines, but data is really what it's about. That's what organizations are really struggling with, and even mature companies still face challenges around governance, security, and regulatory compliance.

The Missing Link Between AI and Cybersecurity

You have AI scientists and engineers who are thinking about getting access to data, deploying models, and improving prediction accuracy. Then you have cybersecurity organizations focused on firewalls, VPNs, and traditional attack vectors. What's not well understood by traditional cybersecurity teams are all the new attack surfaces introduced by AI.
To really understand those attack surfaces, you'd almost have to be an AI scientist. But the AI scientists aren't usually thinking about security. There's a gap between those disciplines that needs to be filled, and that's one of the biggest opportunities for professionals entering this space.

Why Veil Takes a Different Approach to AI Security

With homomorphic encryption, you're typically looking at an inflation factor [of data storage] of about 500 times. If you had a gigabyte of data that needed to move through a system, now it becomes 500 gigabytes. It costs more, takes longer, and for many companies it's simply not practical.
What Veil does is create an irreversible encoding of the data. We preserve exactly the signal the model needs to make predictions while removing the sensitive information an attacker would want. The result is typically more than 95 percent compression while preserving predictive utility. You're not making the trade-offs required by homomorphic encryption or differential privacy.
ABOUT THE AUTHOR
James Maguire
Executive Director
An award-winning journalist, James has held top editorial roles in several leading technology publications, covering enterprise trends in cloud computing, AI, data analytics, cybersecurity and more. He regularly communicates with industry analysts and experts and has interviewed hundreds of technology executives.
Cloudera's Venkat Rajaji: advantages of open source AI
This is some text inside of a div block.
Snowflake's Christian Kleinerman: AI and the speed of data access
This is some text inside of a div block.
David Linthicum: LLM trends
This is some text inside of a div block.
Extreme Networks' Nabil Bukhari: how AI democratizes technology
This is some text inside of a div block.
Komprise's Krishna Subramanian: an AI model for every task
This is some text inside of a div block.
Company
Buyer’s BriefNominate a GuestExecutive SponsorshipVideo PodcastsViewpointsQuick Takes
Quick Link
About UsContact UsSponsor TechVoices
© 2026 Omnibeam, Inc. All rights reserved.
Terms of Service
Privacy Policy
Expert Insight on the Future of Technology

Get in touch with us.

Whether you’re a reader, contributor, or tech enthusiast, we’d love to hear from you. Use the form below to send us a message — we’ll get back to you as soon as we can.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.