Lie Detector for AI
in the Physical World
Reality verification for the external data and AI-derived information physical systems rely on - before hallucinated, poisoned or manipulated information can influence physical action. GIX Intelligence calls this Cyber-Physical Reality Verification (CPRV).
What GIX Intelligence Does
Learns Reality
Learns the target’s reality, dependencies and operating context.
Detects the Lie
Checks external data and AI-derived information against policy.
Enforces Reality
Applies policy outcomes deterministically in hardware.
We Are GIX Intelligence
AI is moving from screens into machines, robots, infrastructure and other physical systems. Once false or manipulated information can influence physical action, cybersecurity must protect not only systems and networks, but the reality those systems trust.
GIX Intelligence is building Cyber-Physical Reality Verification (CPRV): a new cybersecurity layer that works like a lie detector for AI and structured data. The Reality Learning System derives a customer-approved Reality Policy; the T-BOX independently verifies and enforces that policy at runtime.
One system learns reality. One system enforces it.
GIX separates learning from runtime enforcement. The Reality Learning System maps dependencies and derives a candidate Reality Policy. The T-BOX then independently checks incoming external structured data and AI-derived information against that approved reality at the protected boundary.
Learn
The Reality Learning System maps dependencies and derives a candidate Reality Policy.
Prove
Customer-approved, stress-tested through attack simulation, compiled, then burned into the T-BOX.
Shadow
Live events are evaluated without operational impact.
Enforce
The T-BOX Bridge enforces the approved Reality Policy in hardware at the protected boundary.
Learns Reality
Maps target-specific dependencies and derives Reality Signatures from selected external data.
Policy-Driven Enforcement
Evaluates incoming information against the Reality Policy, then enforces the required outcome at the protected boundary.
Purpose-Built Hardware
Purpose-built hardware keeps enforcement independent through a deterministic, unidirectional path with no general-purpose CPU or OS.
The result: AI gets room to operate, your operation gets the efficiency of real-time intelligence, and your CISO gets to say "yes".
One team. Operators and advisors.
Built by people who have run physical operations, secured them, and manufactured for them, guided by an advisory board of active OT CISOs, industry experts, and researchers.
Core Team
Yonathan Cohen
Software architecture. Hardware prototyping. Defence operations and wartime intelligence. Tackles major challenges with minimal resources.
Advisory Board
Benny Ben Sasson
Defence cyber operations and global security leadership, from the defence establishment to global enterprise security (ZIM).
Amir Tsafnat
CISO specializing in government-sector ICS/OT and enterprise IT infrastructure, backed by an M.A. from Ben-Gurion University.
Daniela Zaltsberg
Experienced Project Manager leading multidisciplinary OT/IT deployments and SCADA cybersecurity initiatives for national critical infrastructure.
Ophir Oren
Head of Cyber and AI Security Innovation with over 20 years of expertise in enterprise defense, OT/AI security, and VC advisory.
Prof. Sarit Kraus
Award-winning AI expert and globally recognized pioneer in multi-agent systems, dedicated to building trustworthy human-AI environments.
Moshe-Ishay Cohen
Award-winning Physicist and Technion Ph.D. combining expertise in algorithms and non-linear optical systems for breakthrough deep-tech.
Incoming Team
Currently transitioning from their existing positions to join GIX.
CEO ♂
Deputy CEO ♀
Cyber Physical AI Lab Researcher Lead ♂
Lead Physical AI Models Lab Researcher ♂
Chief develops business success ♀
Questions we get.
What does GIX Intelligence do?
GIX Intelligence builds Cyber-Physical Reality Verification (CPRV). An independent cybersecurity layer that learns the protected target’s unique Reality DNA - its relevant physical reality, dependencies and Reality Signatures, and turns it into a customer-approved Reality Policy. The T-BOX then independently verifies incoming structured data and AI-derived information against that policy before the target relies on it.
Why does this matter now?
AI and real-time external data are moving into the multi-trillion-dollar physical economy. As factories, robots, smart buildings, vehicles and critical infrastructure become more connected and autonomous, the number and importance of digital-to-physical trust boundaries will grow dramatically. Cybersecurity can no longer protect only systems and networks - it must also protect the physical consequences of what those systems believe and do.
Why isn’t existing cybersecurity enough?
Existing controls solve different parts of the problem. Firewalls control access and traffic. Data diodes enforce direction. AI-security products protect models and applications. GIX addresses a different question: does the incoming external information match the customer-approved physical reality of the specific target before that target relies on it?
Who is it for?
Any organization where data from the internet, another network or an AI system can influence a physical outcome: from factories and robots to smart buildings, mobility, data centers, defence and critical infrastructure. GIX matters most where false, poisoned or misleading information can cost millions, disrupt operations or put lives at risk, and where trusted real-time external data can unlock equally significant value.
Does it require changing our systems?
GIX is designed for simple deployment at a defined cyber-physical trust boundary - without rewriting applications or replacing the systems you already trust. It adds an independent Reality Verification layer at the boundary where external data and AI-derived information enter the physical system.
How do we start?
Show us one place where external AI or data can influence a physical system. We map the trust boundary, learn its Reality DNA, build the Reality Policy with you, and integrate GIX into the existing data path.
Does GIX require full payload decryption?
No. GIX does not require blanket decryption of your traffic. Enforcement can operate on metadata, timing, and cross-system context at the protected boundary; where payload visibility is required, it is mapped explicitly and minimally during deployment.
Does the T-BOX Bridge depend on the Reality Learning System at runtime?
No. Once the Reality Policy is approved, the T-BOX Bridge enforces it independently at runtime. The data it evaluates can still come from approved external sources, including cloud services or other networks.
Start the conversation.
Tell us where AI, automation or external data can affect a physical system. We’ll help define the trust boundary and unlock the value of real-time data safely.