Organizations
Organizations that develop a listed implementation, publish key work on a mechanism, or run infrastructure for the field.
Companies (13)
Amodo Design
A hardware engineering company that builds and publishes prototypes for verifying AI agreements, including inference recomputation, network taps and memory wiping.
9 related records · 15 publications
Related records
- Sampled inference recomputation
- Bandwidth limits and compartmentalization
- Timed challenge-response and memory-occupation challenges
- Workload classification from telemetry and side channels
- Memory wiping and proofs of secure erasure
- Network taps and certifiers
- Whole-workload recomputation (reproducible packets)
- DiFR (Divergence From Reference)
- AI 2040 inference-only verification stack
Publications
- AI 2040 Plan A — Verification SITREP
- Amodo-Design/Inference-Recomputation-Prototype (GitHub repository)
- Amodo-Design/PoSE-Memory-Wiping (GitHub repository)
- An Inference Verification Prototype — Stage 1
- Example Schemes for Verifying High-Stakes AI Agreements
- Fitting a Network TAP to our Inference Verification Prototype
- Improving Disk Wiping Speed for Memory Wipes
- Memory Wipes - Performance Analysis
- Network Tapping for AI Verification: A Technical Assessment
- Network Taps — A First Test
- Network Traffic Hashing
- Scaling Recomputation Inference Verification
- The Tray as a Bandwidth Boundary
- Understanding Data Center Power Delivery
- Verifying international AI deals: Plan A, the state-of-play, and what you can do to help
Apple
A company; developer of Private Cloud Compute, the server system for Apple Intelligence, where devices send requests only to servers attesting to publicly logged software.
1 implementation · 4 publications
Implementations
Publications
Attestable
A company developing zero-knowledge proofs for AI systems; it reports a prover for large language model inference and proposes proof-based compute accounting.
1 implementation · 2 related records · 3 publications
Implementations
Related records
Publications
Gensyn
A company building AI forecasting systems; developer of Verde, a protocol for checking delegated machine-learning jobs, and the Reproducible Execution Environment (REE) for model inference.
1 implementation · 7 publications
Implementations
Publications
- Building Delphi: Pricing, Settlement, and Agentic Trading
- gensyn-ai/ree: Gensyn Reproducible Execution Environment (GitHub repository)
- Reproducible Execution Environment (REE) (Gensyn documentation)
- What is Delphi? (Delphi documentation)
- Introducing Judge
- Verde Verification System In Production
- Verde: Verification via Refereed Delegation for Machine Learning Programs
Lucid Computing
A company offering attested, confidential-computing AI clusters; it hosts the draft Sovereignty Certificates specification for ping-based chip location verification and runs a verification research programme.
1 implementation · 2 related records · 6 publications
Implementations
Related records
Publications
Meta
A company; developer of Private Processing, the server system for WhatsApp AI features, where clients check confidential virtual machine attestations before connecting.
1 implementation · 1 publication
Implementations
Publications
NVIDIA
An accelerated-computing company whose Hopper and Blackwell data-centre GPUs offer a confidential-computing mode with hardware attestation.
2 related records · 9 publications
Related records
Publications
- Introducing NVIDIA Fleet Intelligence for Real-Time GPU Fleet Visibility and Optimization
- Meta Builds AI Infrastructure With NVIDIA
- NVIDIA Trusted Computing Solutions Release Notes (R595 TRD1)
- Boost GPU Memory Performance with No Code Changes Using NVIDIA CUDA MPS
- No Backdoors. No Kill Switches. No Spyware.
- NVIDIA Secure AI with Blackwell and Hopper GPUs (White Paper)
- Opt-In NVIDIA Software Enables Data Center Fleet Management
- Now in General Availability: NVIDIA H100 GPUs in Microsoft Azure Confidential Virtual Machines
- Confidential Computing on NVIDIA H100 GPUs for Secure and Trustworthy AI
Pearl Research Labs
A company building AI inference infrastructure and Pearl, a blockchain designed to make proof-of-useful-work mining a by-product of GPU matrix multiplication.
1 implementation · 1 related record · 5 publications
Implementations
Related records
Publications
Planet Labs
A company that builds and operates Earth-imaging satellites and sells their imagery; it is one commercial source used to track AI data-centre construction.
1 related record · 1 publication
Related records
Publications
Prime Intellect
An AI compute, training and inference company; developer of TOPLOC, a hashing scheme for verifying LLM inference.
1 implementation · 1 related record · 6 publications
Implementations
Related records
Publications
- INTELLECT-2: A Reasoning Model Trained Through Globally Decentralized Reinforcement Learning
- PrimeIntellect-ai/toploc (GitHub repository)
- SYNTHETIC-2
- SYNTHETIC-2 Release: Four Million Collaboratively Generated Reasoning Traces
- TOPLOC: A Locality Sensitive Hashing Scheme for Trustless Verifiable Inference
- TOPLOC: A Locality Sensitive Hashing Scheme for Trustless Verifiable Inference (blog post)
Thinking Machines Lab
An AI research and product company; developer of batch-invariant kernels that make language-model outputs independent of batch size.
1 implementation · 2 publications
Implementations
Publications
Tinfoil
A company offering AI inference inside secure hardware enclaves, with remote attestation that clients can check; developer of the Modelwrap model-identity tool.
1 implementation · 4 related records · 7 publications
Implementations
Related records
- Hardware-attested weight binding
- TEE remote attestation for AI workloads
- Confidential multi-party verification
- Safeguard attestation
Publications
Research organizations (8)
AI Futures Project
A small research group forecasting the future of AI; publisher of the AI 2040 scenario, including an inference-only verification plan.
1 implementation · 7 related records · 3 publications
Implementations
Related records
- Sampled inference recomputation
- Bandwidth limits and compartmentalization
- Chip registries and manufacturing records
- Memory wiping and proofs of secure erasure
- Network taps and certifiers
- Remote detection of data centres
- Whole-workload recomputation (reproducible packets)
Publications
Center for a New American Security
A nonprofit national-security policy organization; publisher of a report proposing on-chip mechanisms, including location verification, for governing AI chips.
3 related records · 2 publications
Related records
- Chip location verification
- Hardware performance throttling and licensing
- Hardware-enabled guarantees (flexHEG) and guarantee processors
Publications
Centre for the Governance of AI
An AI governance think tank; its papers analyse how compute can be used to govern AI and propose certifying state jurisdictions against international oversight standards.
Epoch AI
A research institute studying the trajectory of AI; its AI data centers hub estimates AI data-centre capacity from satellite imagery and permits.
1 related record · 2 publications
Related records
Publications
Institute for AI Policy and Strategy
A nonpartisan think tank on AI policy; its work on AI chips includes a design for delay-based location verification and a review of export-verification methods.
2 related records · 4 publications
Related records
Publications
Intelligence Security Laboratories
A nonprofit research lab that aims to develop and demonstrate high-security AI systems and to integrate security across whole data centres.
Machine Intelligence Research Institute
A nonprofit working to help prevent human extinction from artificial superintelligence; its Technical Governance Team publishes designs and analyses for verifying AI agreements.
2 implementations · 13 related records · 7 publications
Implementations
Related records
- Hardware-attested weight binding
- Sampled inference recomputation
- TEE remote attestation for AI workloads
- Bandwidth limits and compartmentalization
- Confidential multi-party verification
- On-chip telemetry from timing, memory and performance counters
- Safeguard attestation
- Timed challenge-response and memory-occupation challenges
- Workload classification from telemetry and side channels
- Memory wiping and proofs of secure erasure
- Network taps and certifiers
- Proofs of useful work for capacity accounting
- Side-channel suppression for isolated facilities
Publications
- A System Overview for Near-Term, Low-Trust AI Compute Verification
- De-risking Interconnect Limits for AI Verification
- On TEEs for Privacy-Preserving Monitoring in AI Governance
- Summary: TGT's 2026 ICML Papers
- Suppressing Side Channels in an Untrusted Data Center via Retrofitted Defenses
- An International Agreement to Prevent the Premature Creation of Artificial Superintelligence
- Mechanisms to Verify International Agreements About AI Development
RAND
A nonprofit, nonpartisan research organization; its reports cover verification of international AI agreements, hardware-enabled governance mechanisms and secure inference data centres.
1 implementation · 3 related records · 4 publications
Implementations
Related records
- Chip registries and manufacturing records
- Hardware performance throttling and licensing
- Hardware-enabled guarantees (flexHEG) and guarantee processors
Publications
- Highly Secure Inference Data Centers: A Vertically Integrated Strategy for Security Engineering
- Verifying International Agreements on AI: Six Layers of Verification for Rules on Large-Scale AI Development and Deployment
- Hardware-Enabled Governance Mechanisms: Developing Technical Solutions to Exempt Items Otherwise Classified Under Export Control Classification Numbers 3A090 and 4A090
- Securing AI Model Weights: Preventing Theft and Misuse of Frontier Models
Nonprofits (4)
Future of Life Institute
A nonprofit working on extreme risks from transformative technology; it built secure-hardware proofs of concept for AI governance and partners on a network-logger prototype.
2 related records · 2 publications
Related records
Publications
OpenMined
A nonprofit developing PySyft, software for joint analysis of private data, including confidential AI evaluations.
1 implementation · 1 related record · 2 publications
Implementations
Related records
Publications
Singapore AI Safety Hub (SASH)
A Singapore-based AI safety research and field-building organization; with international partners, it develops a confidential network logger for inference verification.
1 implementation · 1 related record · 3 publications
Implementations
Related records
Publications
Verifiable Compute Foundation
An independent nonprofit providing free bare-metal GPU access and physical instrumentation for AI verification research.
Universities and labs (4)
Hardware AI Governance Lab
An Oxford lab studying hardware for AI governance; its papers propose fingerprinting all data entering or leaving AI clusters and analyse verification of semiconductor manufacturing.
1 related record · 2 publications
Related records
Publications
Oxford Martin AI Governance Initiative
A research programme at the University of Oxford that studies AI governance from technical and policy angles and hosts the Hardware AI Governance Lab.
1 implementation · 5 publications
Implementations
Publications
- Verifiable Semiconductor Manufacturing
- Verification for International AI Governance
- Computing Power and the Governance of Artificial Intelligence
- Governing Through the Cloud: The Intermediary Role of Compute Providers in AI Regulation
- International Governance of Civilian AI: A Jurisdictional Certification Approach
University of Cambridge
A UK university; researchers in its Department of Computer Science and Technology developed Attestable Audits, AI benchmarks run and attested inside trusted execution environments.
1 implementation · 4 related records · 1 publication
Implementations
Related records
- Hardware-attested weight binding
- TEE remote attestation for AI workloads
- Confidential multi-party verification
- Safeguard attestation
Publications
University of Waterloo
A university in Waterloo, Ontario, Canada, whose researchers developed zkLLM, a zero-knowledge proof system for large language model inference.
2 implementations · 1 related record · 2 publications
Implementations
Related records
Publications
Consortia (1)
Confidential Computing Consortium
A Linux Foundation project on confidential computing whose special interest groups work on attestation interoperability and attestation-based workload identity.