Key takeaways
The June 22, 2026 executive order, Securing the Nation Against Advanced Cryptographic Attacks, requires federal agencies to migrate their most important systems to post-quantum cryptography, with key establishment due by December 31, 2030 and digital signatures by December 31, 2031.
National Security Presidential Memorandum/NSPM-11, signed June 5, 2026, requires that AI systems used in national security be controllable, reliable, and resistant to unauthorized modification, and cannot be disabled or materially altered without federal approval, with test and verification methods due within 120 days.
These should not be two separate compliance projects. Satisfying both durably converges on the same foundation: records that are quantum-resistant and independently verifiable from the moment they are created.
Encryption can be retrofitted to systems you already run. A verifiable record cannot. Provenance has to be built in when the record is created, which is what makes the AI buildout happening now the deciding window.
Intrana is a quantum-resistant distributed ledger built for both requirements at once, using NIST-standardised algorithms, with verifiable, tamper-evident records as a native property.
On June 22, 2026, President Trump signed the executive order Securing the Nation Against Advanced Cryptographic Attacks. It put hard dates on something the federal government had been circling for years. Agencies must move high-value assets and high-impact systems to post-quantum cryptography for key establishment by the end of 2030, and for digital signatures by the end of 2031. The order also directs that federal procurement and contracting rules follow.
Two and a half weeks earlier, on June 5, the administration issued NSPM-11, "Artificial Intelligence in the National Security Enterprise." Most of that memorandum is about accelerating AI adoption across defense and intelligence. Two parts of it are not. Its Assurance pillar requires that a fielded AI system cannot be disabled, degraded, or materially modified without federal knowledge and approval, and it directs standardised testing, evaluation, verification, and validation methods within 120 days. Its Accountability pillar requires a human chain of command running from the American people to the warfighter, with commanders responsible for how AI is used in the field. Without tamper-evident records of what an AI system did, that chain of command has no reliable foundation on which to account for decisions after the fact.
The default reaction in most organizations is to treat these as two separate workstreams: a cryptography migration on one side, an AI governance program on the other. Teams that split them that way will go through two separate due diligence and procurement processes, spend twice and still leave seams because they’re not looking at it as an infrastructure problem. The clearer framing is that satisfying both converges on the same architectural foundation.
Here is the asymmetry that matters. Encryption can be added to a system you already run. You can inventory what you have, prioritize it, and migrate it on a schedule, which is exactly what the post-quantum order tells agencies to do. A verifiable record does not work that way. You cannot make a record trustworthy after the fact. The proof of where a piece of data came from, which model produced it, and whether it was altered has to be captured at the moment the record is created. If it was not, no later upgrade can retrofit provenance.
That is why the AI side is the harder problem. The systems generating the records that will matter most—training datasets, model inferences, model versions, sensor readings, autonomous actions—are being stood up right now. Every one of those records may later have to be defended, in an after-action review, an audit, a contract dispute, or a courtroom, against an adversary with an interest in claiming it was forged or tampered with. The window to build provenance into a record closes the moment that record is created. The records accumulating right now are either cryptographically provable or they are assertions, and for those specific records, each outcome is permanent.
The post-quantum cryptography Executive Order: hard deadlines, named threat
The executive order, Securing the Nation Against Advanced Cryptographic Attacks, requires federal agencies to move their most important systems to post-quantum cryptography by these deadlines: key establishment by December 31, 2030, and digital signatures by December 31, 2031. Agencies have to designate a migration lead within 30 days, inventory their cryptography, and build a prioritized plan under guidance from the Office of Management and Budget due within 90 days. NIST runs a migration pilot to be completed by the end of 2027, and CISA, with NIST, is to publish guidance for a cryptographic bill of materials, essentially an ingredient list of the cryptographic standards inside a given system.
The order also names the threat directly. Adversaries may already be collecting encrypted federal data today, intending to decrypt it once a large enough quantum computer exists. Security teams call this “harvest now, decrypt later.” It is the documented operating assumption, not hypothetical. Data with a long confidential life (e.g., diplomatic records, health records, defense archives) is the most exposed.
This is the clear half of the picture. The standards exist, the deadlines exist, and the first step, a cryptographic inventory, is close to the same for everyone.
NSPM-11 and the verifiability problem most teams will underestimate
NSPM-11 requires that AI systems used in national security be controllable, reliable, and resistant to unauthorized modification, and that they cannot be disabled or materially altered without the government's knowledge and approval. It is easy to mischaracterize otherwise, so it is worth being precise. The memorandum is built around four pillars—Adoption, Adaptation, Assurance, and Accountability—and the bulk of it is about getting advanced AI into the hands of national security users faster. The verifiability requirement lives primarily in the Assurance pillar, and at this stage it is a direction rather than a finished technical standard. The TEVV methodologies (Test, Evaluation, Verification, and Validation) are due within 120 days, and they will set the contract language and program requirements vendors actually have to meet later.
Accountability is the second hook, and the one that tends to get less attention. The memorandum requires a human chain of command responsible for how AI is used in the field. That sounds organizational, and it is. But a commander cannot meaningfully account for a decision they cannot trace with confidence. If a congressional inquiry or inspector general asks what the model did, what data it used, and whether any of it was altered, the answer requires a tamper-evident record that predates the question. A command structure without post-quantum cryptographic proof beneath it is not truly accountable.
As the Intrana for Earth whitepaper frames the stakes: “A world in which artificial intelligence systems generate, ingest, and act upon digital artifacts is a world in which the cryptographic substrate beneath those artifacts becomes load-bearing for epistemology itself.”
That is a verifiability requirement, and verifiability is not something an encryption migration delivers. Confidentiality and integrity are different properties. You can encrypt a model's outputs and still have no independent way to prove they were not swapped. The order points at a gap that the post-quantum work, on its own, does not close.
Why the two mandates converge on one foundation
Post-quantum cryptography and AI verifiability are different mandates with different authors and different deadlines. An organization that wants to satisfy both needs a cryptographic substrate that produces tamper-evident, independently verifiable records as a native property rather than as a feature bolted on afterward.
Solving them separately is where the cost and the risk come from. A standalone cryptography migration handles confidentiality but leaves verifiability unaddressed. A separate AI governance platform layered onto legacy infrastructure inherits whatever trust gaps that infrastructure already has, and adds integration seams an adversary can work between. The two tracks duplicate effort and still do not meet in the middle.
This is the case for treating the substrate itself as the unit of work.
Where Intrana fits
Intrana is a quantum-resistant distributed ledger, and, to our knowledge, the only production-grade substrate that simultaneously addresses the cryptographic obsolescence horizon and the verifiability requirements of large-scale AI deployment. The design choices behind that claim map onto the two requirements as follows.
On the cryptography, Intrana was built quantum-resistant from the genesis block rather than planned for a future migration. Every transaction carries a hybrid signature: currently classical Ed25519 paired with NIST-standardised CRYSTALS-Dilithium (ML-DSA), so a transaction stays secure as long as either scheme holds. The default is ML-DSA-65, with ML-DSA-87 available for the highest-assurance applications, which is the same level-5 parameter set that the NSA's CNSA 2.0 suite mandates for national security systems. Key exchange uses CRYSTALS-Kyber (ML-KEM), hashing uses BLAKE3 at 512-bit output, and the architecture supports SLH-DSA, HQC, and future algorithms as alternatives for cryptographic agility. These are the FIPS-standardised families the executive order points agencies toward, with integration happening at the API level.
On verifiability, the ledger produces tamper-evident, independently verifiable records by design. That extends to the artifacts NSPM-11 cares about: cryptographic binding and provenance for training data, model versions, and inferences, so that what a model was and what it produced can be checked later by a party that did not generate it. Zero-knowledge proofs let a computation be verified without exposing the underlying data, which matters when a record is sensitive but still has to be provable. A three-tier validator structure, with entropy-based pairing and Byzantine fault tolerance, provides institutional accountability and deterministic finality.
A note on performance, since it tends to come up. The architecture is designed to sustain on the order of 5,000 transactions per second with one-second block times and immediate finality. The relevant point here is that the verifiability sits on the same path that produces the throughput, rather than being added as a separate audit step after the fact.
None of this has to be taken on faith. The reference implementation is open source under a permissive license, with SDKs and a testnet, so a technical team can exercise the cryptography and the verifiability directly before committing to anything.
Who should be moving now
The orders apply most directly to federal agencies, defense and intelligence contractors, and critical infrastructure operators, but the underlying exposure reaches further. The clearest circle:
Federal agencies and national security organizations directly pursuant to the post-quantum order and NSPM-11.
Prime contractors and subcontractors in the defense and intelligence supply chain, who face procurement requirements as they reach new contracts and will inherit AI assurance terms over time.
Critical infrastructure operators in energy, finance, healthcare, water, and transportation, who carry both the migration burden and a growing expectation of verifiable AI in their operations.
Financial institutions that need quantum-safe settlement and a defensible record of AI-driven decisions.
Any organization holding data with a long confidential life, or running production AI whose outputs may later be challenged.
If your records have to remain trustworthy for decades, or have to stand up to an adversary who benefits from disputing them, you are in this circle whether or not a regulation names you yet.
The honest case for starting early
Starting now is the lower-risk choice, even though the deadlines are years away, for two reasons.
The first is the cryptographic inventory. Almost every plan starts there, it takes longer than teams expect, and the order effectively requires it. Organizations that have not started one are already behind the agencies that have.
The second is the part you cannot buy back later. The records AI systems generate today are accumulating now. If provenance was not built in when they were created, a future migration will not add it. That is the real difference between the two halves of this problem. The cryptography you can schedule. For records that exist without provenance, no future migration adds it.
The lower-risk way to approach these compliance requirements is to integrate infrastructure that makes your data both quantum-resistant and verifiable at the same time. Intrana is the solution.
Frequently asked questions
What is harvest now, decrypt later?
Harvest now, decrypt later is a strategy where an adversary collects encrypted data today and stores it to decrypt later, once a large enough quantum computer exists. It matters for any data whose sensitivity outlasts the arrival of quantum computers, because that data is effectively exposed the moment it is captured, even though the decryption is years away. The June 2026 executive order names this threat directly.
What are the NIST post-quantum standards?
The NIST post-quantum standards are three FIPS publications finalized in August 2024: FIPS 203 (ML-KEM, for key establishment), FIPS 204 (ML-DSA, for digital signatures), and FIPS 205 (SLH-DSA, a hash-based signature alternative). They are the algorithms the executive order directs federal systems to adopt. Intrana uses ML-DSA and ML-KEM, the families standardised in FIPS 204 and FIPS 203.
What is CNSA 2.0, and is Intrana aligned with it?
CNSA 2.0 is the NSA's Commercial National Security Algorithm Suite 2.0, the post-quantum standard for national security systems. It mandates the highest parameter levels, ML-KEM-1024 and ML-DSA-87, with a procurement gate that begins January 1, 2027. Intrana implements those families and supports the level-5 ML-DSA-87 parameter set CNSA 2.0 specifies. Formal CNSA 2.0 compliance for a national security system also requires NSA or NIAP validation, which is a separate certification process.
Can AI verifiability be added to systems after the fact?
Not retroactively. You can add verifiable record-keeping to a system going forward, and records created from that point will carry provenance. What you cannot do is prove the integrity of records that were created before provenance was in place. Encryption can be layered onto a system you already run. The cryptographic proof of what a record was at the moment of its creation cannot be added after the fact. The records worth protecting are the ones being generated now.
What is the difference between encryption and verifiability?
Encryption protects confidentiality: it keeps data secret. Verifiability protects integrity and provenance: it proves where a record came from, what produced it, and whether it was altered. They are different properties. You can encrypt a record and still have no independent way to prove it was not swapped or forged, which is why a post-quantum encryption migration does not, on its own, satisfy a verifiability requirement.
How long does a first deployment take?
A first deployment is not a full enterprise migration. It is a single critical system, evaluated end to end, on infrastructure that is quantum-resistant and verifiable from the first record. Scoped that way, it is an exercise measured in weeks rather than years, and it produces something concrete to assess before any larger commitment.
Does the post-quantum requirement reach federal contractors?
The hard deadlines in the executive order fall on federal agencies. The order also directs procurement and contracting rules to follow, so contractors should expect post-quantum requirements to flow down through acquisition over time, in the same pattern as existing federal cybersecurity rules. For national security systems specifically, the CNSA 2.0 procurement gate already begins January 1, 2027.
Where to go next
The technical foundation is documented in Part VII of the "Intrana for Earth" whitepaper. If you want to see how this maps to a specific sector (e.g., defense, finance, healthcare, energy, or government) that is the best place to start, and a single-system evaluation is the most direct way to test whether the claims hold in your environment.
Sources
The White House, "Securing the Nation Against Advanced Cryptographic Attacks," June 22, 2026:
NSPM-11, "Artificial Intelligence in the National Security Enterprise," June 5, 2026:
NIST FIPS 203, Module-Lattice-Based Key-Encapsulation Mechanism Standard (ML-KEM):
https://csrc.nist.gov/pubs/fips/203/final
NIST FIPS 204, Module-Lattice-Based Digital Signature Standard (ML-DSA):
https://csrc.nist.gov/pubs/fips/204/final
NIST FIPS 205, Stateless Hash-Based Digital Signature Standard (SLH-DSA):
https://csrc.nist.gov/pubs/fips/205/final
CISA, Post-Quantum Cryptography Initiative:
Intrana: A Quantum-Resistant Substrate for Verifiable Civilization-Scale Infrastructure: