Dr. Jonathan Valamehr (Jonny), INT31
Foundational technologies demand an uncompromising approach to security, due to their critical role in safeguarding root of trust. For Confidential Computing technologies such as Intel® Trust Domain Extensions (Intel® TDX), the goal is to protect sensitive workloads, even against compromised hypervisors or malicious insiders for billions of users. Intel TDX achieves this through the enablement of Confidential Virtual Machines (CVMs) — also known as Trust Domains (TDs) — that provide strong hardware-enforced confidentiality and integrity guarantees in multi-tenant and cloud environments.
Delivering on these commitments requires more than innovative hardware features. Intel integrates security assurance discipline throughout product design, development and support using a rigorous Security Development Lifecycle (SDL) that spans hardware, firmware, and software. For Intel TDX, this includes a comprehensive set of security-focused activities, such as:
- Early security architecture and threat model reviews to validate security objectives and assumptions before designs are finalized
- Design and code analysis across hardware IP, firmware, and software components
- Continuous engagement with engineering teams to identify, track, and mitigate risks well before product release
Figure 1: Intel’s Security Development Lifecycle integrates security across six phases.
Additionally, Intel applies Offensive Security Research (OSR) techniques to evaluate Intel TDX from an attacker’s perspective. This work includes:
- Adversarial threat modeling and architectural gap analysis
- Vulnerability research across hardware, firmware, and software layers
- Platform-level security testing in realistic deployment environments
- Focused security hackathons designed to uncover complex, cross-layer attack scenarios
This ensures that Intel TDX is continuously challenged, validated, and hardened against real-world threats.
At the same time, it’s understood that no product can be absolutely secure. The more eyes that look at product, the stronger it will be. That’s why Intel works closely with our partners, combining expertise, security capabilities, and use case knowledge to continually enhance technology.
Today, results of a recent collaboration with Google on Intel TDX were published, showcasing tremendous work to strengthen this confidential computing solution.
Bringing Together Experts from Two Industry Leaders
Members of the INT31 research team in conjunction with the Intel TDX Security Research team conducted this joint security review with Google engineers as an extension of previous research performed in 2023. The Google Cloud Security team worked closely with INT31 over five months in 2025 to identify potential vulnerabilities, bugs, and other design considerations in the Intel TDX Module 1.5 code that is responsible for the high-level functions of TDX.
Focus centered on two features: Live Migration, which allows a TD to move from one host platform to another while still executing, and TD Partitioning, or partitioned Virtual Machines within TDs. The Intel TDX Module 1.5 version that Google reviewed was public at the time of this engagement.
How the Joint Research was Done
During the review, Google used a number of methods to identify security issues in the Intel TDX module code:
- Manual code review
- Development of their own tools to help automate bug-finding and proof-of-concept development
- Off-the-shelf AI tools (particularly Gemini 2.5 Pro and NotebookLM)
INT31 team members provided technical guidance when necessary, triaged findings to the appropriate Intel feature owners for investigation and to drive mitigation, and provided a technical advisory to customers about the findings.
What the Research Found
Using the methods above, Google found five vulnerabilities in Intel TDX Module 1.5 code and flagged another 35 weaknesses, bugs, and security improvement suggestions. All are outlined in their report which details their journey, methods, and results used during this engagement.
At the time of this publication, all five of the vulnerabilities have been patched in the most recent version of Intel TDX Module code which has been released to partners. This research illustrates how Intel is committed to our Security First pledge. Customers can take confidence that it’s not just Intel working to strengthen our technology, but the ecosystem working together to raise the security bar.
“We greatly appreciate the collaboration with Google on the TDX 1.5 security review as it not only demonstrated how industry can come together to strengthen and project trust in the most important security technologies underlying our digital infrastructure, but also that AI can be used at scale to perform security assurance on complex products like Intel TDX. Intel is reviewing how we can put many of these new techniques into practice in our future product assurance processes.” – Simon Johnson, Intel Fellow, Confidential Compute
Intel and the INT31 team continue to hold our products to the highest levels of security assurance, leveraging best practices, internal red teaming, and external collaboration to ensure foundational technologies continue to safeguard customer data in the wild. We are very proud of this work and the levels of security we have built into Intel TDX. Google’s continued collaboration on Intel TDX helps battle-test our technology and by extension makes Google’s cloud offerings even more secure for their customers.
"Our deep collaboration with Intel allows us to battle-test and strengthen the security of foundational technologies that power Confidential Computing. By proactively identifying vulnerabilities in critical features like Live Migration and TD Partitioning using advanced AI tools like Gemini, we are helping to raise the security bar for the entire ecosystem." - Andrés Lagar-Cavilla, Distinguished Engineer, Google Cloud
Share Your Feedback
We want to hear from you. Send comments, questions, and feedback to the INT31 team.
Acknowledgements
This collaboration reinforces a key principle: securing foundational technologies is not a one-time effort, nor is it something any single organization can do in isolation. Intel and the INT31 team are immensely grateful to Kirk Swidowski, Josh Eads, Daniel Moghimi, and Erdem Aktas from Google for reporting these issues. We look forward to a continued collaboration on Intel TDX and other critical technologies.
More Information
- Read the Google technical paper
- Read the Google Bughunters blog
- Watch the Intel Chips & Salsa video
- Read the Intel security advisory
- Learn more about Intel Product Security Assurance
About the Author
Dr. Jonathan Valamehr (Jonny) is a Director and Principal Engineer in the INT31 Security Research team within the Intel Product Assurance and Security (IPAS) group at Intel. His team focuses on cryptography, offensive security research, assurance, and threat modeling. Prior to joining Intel, Dr. Valamehr co-founded Cycuity, a semiconductor cybersecurity assurance company. He has 16 publications and 3 patents in the areas of computer architecture and hardware security, and received his Ph.D. in Computer Engineering from UC Santa Barbara.