🔍 Read the full analysis: Finance, Defence And The Emerging AI Cryptography Challenge on ThorstenMeyerAI.com
Prime made for students and young adults
- Fast, free delivery for dorm and study essentials
- Prime Video and Amazon Music included
- Member-only deals
TL;DR
OpenAI published 722 mathematical manuscripts produced with an internal AI model, including claims involving computational complexity. The release and warnings from cryptocurrency figures have renewed debate about whether AI could uncover algorithms that weaken cryptographic systems, including post-quantum standards. No cryptographic protocol has been shown to be broken, and the manuscripts require expert verification.
OpenAI published 722 mathematical manuscripts on October 6, generated by an unreleased internal model, prompting new discussion about whether AI could find algorithms that weaken cryptographic systems used in finance, intelligence and defence. The publication reports no cryptographic break: the manuscripts contain claims that require checking, and the extent of any practical threat remains unknown.
The manuscripts were grouped into 372 families and reportedly drawn from roughly 4,000 problems. The source says the work used, on average, about three hours of ChatGPT Pro compute per result. Among the claims are results concerning the Unique Games Conjecture, Hilbert’s tenth problem over the rationals and the Riemann zeta function. These remain mathematical claims, not automatically verified discoveries.
For cryptography, the attention has focused on reports of improvements to algorithms for tasks such as integer multiplication and Fourier transforms. A separate result on 3SUM, attributed to researchers Virginia Vassilevska Williams and Josh Alman, reportedly used an idea from an Anthropic model. The source says computer scientist Scott Aaronson noted that cryptography itself was absent from the 722 manuscripts, while reporting that AI companies have begun discreetly testing whether their models can attack important protocols. That testing is not described in enough detail to establish what was tried or with what result.
There has already been at least one reported correction: OpenAI withdrew a claimed proof concerning the Hodge conjecture for products of K3 surfaces, reportedly because of a sign error. This illustrates the distinction between AI-generated mathematical work and results that survive independent review. The source does not report a demonstrated attack on RSA, elliptic-curve cryptography or post-quantum standards.
The old map is gone: AI mathematics, quantum computers and the cryptography holding up finance and defence
For a decade the plan was simple: elliptic curves doomed by quantum; lattices safe; hashes safe. Nothing has been broken. But a second threat has arrived that doesn’t respect those borders — AI producing new mathematics faster than any human community, against assumptions that are believed, not proven.
Now: on borrowed time — possibly shorter than the quantum countdown suggests.
Now: unproven against AI — and the destination most of the world is migrating to.
Now: reminded estimates move — BSI advised against new deployments on 1 Oct 2026.
Now: safest ground available — not a guarantee.
~n log0.9999999999999 n — a barrier many thought fundamental (OpenAI, claimed)
Overturns a half-century conjecture. Williams & Alman; key idea from an Anthropic model
“Conspicuous by its absence” (Aaronson) — labs reportedly testing crypto “gingerly and discreetly”
ECDSA could break before Q-day, “in the worst case in months not years.” Move funds to never-signed addresses. ~6M BTC sit behind exposed keys.
The new risk is the destination of the migration. Hash-only where possible; “much more paranoid” lattice params; ×10 key sizes long-term. Doesn’t recommend anyone scramble.
“No evidence whatsoever” that elliptic-curve assumptions are close to failing.
Classical breaks could reach “quantum-safe” schemes — but don’t treat a two-year scenario as a date.
Known to IBM and the NSA designing DES (~1974); public via Biham & Shamir (~1990); confirmed by Coppersmith (1994).
Invented at GCHQ — RSA- and Diffie–Hellman-equivalents — and kept secret for over two decades.
No crypto in 722 manuscripts. Found and withheld? Not posed? Posed and failed? Indistinguishable from outside.
Traffic recorded today is decrypted when a break arrives. For secrets that must last 25+ years, a break in 2035 is a break today. A state that finds one won’t announce it — it will mine its archives.
Signatures can be built from hashes. Encryption and key exchange need a trapdoor with structure — lattices, codes or group theory. Defence can only choose which structure, how much margin, how many combined.
Every date was set against quantum hardware forecasts with visible warning. The AI threat offers none.
“ML-KEM everywhere” means starting over if lattices weaken. “We can swap algorithms” doesn’t.
Blockchains show a classical break first — exposed keys and balances are public. Monitor dormant exposed addresses.
Every algorithm, key, certificate, protocol.
PQ + classical, as BSI requires.
Firmware, updates, long-term keys.
Highest sets; evaluate FrodoKEM.
More than one mathematical family; HQC coming.
Swap algorithms without rebuilding.
Forward secrecy, rotation, hidden keys.
Buterin: lost more in botched migrations than in all hacks.
Nothing has been broken, and the sceptics are right that there’s no evidence elliptic curves or lattices are about to fall. But the map has changed: elliptic curves on borrowed time, lattices unproven against AI, codes reminded that estimates move, hashes the safest ground available. For finance, intelligence and defence the answer is the same whichever threat arrives first.The quantum threat comes with a countdown. The AI threat may arrive as a silence — an empty folder where a paper should have been. The winners will be those who can change their algorithms fastest.
Why Cryptographic Assumptions Matter
Modern digital systems rely on mathematical problems believed to be difficult to solve with available computing methods. Cryptography protects bank transactions, confidential communications, software updates and military systems. If a new algorithm made a supposedly hard problem tractable, the consequence could reach far beyond cryptocurrency. But a theoretical improvement is not the same as a practical attack: it must be independently checked, translated into a method that applies to a deployed system, and shown to work at relevant scale.
The concern raised here differs from the established quantum-computing threat. A sufficiently capable, error-corrected quantum computer running Shor’s algorithm could break RSA and elliptic-curve cryptography. Governments and companies can monitor visible hardware progress, though its timing remains uncertain. An algorithmic discovery could be harder to observe: it might run on ordinary computers and remain undisclosed by whoever finds it. That possibility is a risk scenario, not evidence that a secret breakthrough exists.
For finance and defence organizations, the immediate relevance is planning rather than panic. They need to understand which systems depend on which mathematical assumptions, and how quickly they can replace cryptographic components if evidence changes. The same applies to intelligence agencies, where confidential information may need protection for years. The source offers no evidence that any institution has changed its migration schedule in response to the manuscript release.
Popular right now
Apple AirPods Pro 3 Wireless Earbuds with Active Noise Cancellation
Vitamix Propel Series 510 Professional-Grade Blender, 3 Programs, Black
Jura E6 Automatic Espresso Machine, Piano Black (15622)
As an affiliate, we earn on qualifying purchases.
From Quantum Migration to AI
For years, post-quantum planning has focused on replacing public-key systems vulnerable to quantum attacks. In August 2024, the U.S. National Institute of Standards and Technology standardized ML-KEM for establishing encryption keys, ML-DSA for digital signatures and SLH-DSA, a hash-based signature standard. The first two are lattice-based; the third relies on hash functions.
The source argues that AI raises questions about whether new algorithms could challenge assumptions behind systems beyond RSA and elliptic curves, potentially including lattice-based cryptography. It also says hash-based systems may be less exposed to this particular concern, but provides no proof that they are immune. The standards’ adoption and security assessments remain separate from the claims in the AI manuscripts.
Cryptocurrency drew early attention because some blockchain addresses publicly expose cryptographic keys after use. On October 7, Ethereum Foundation researcher Justin Drake urged the industry to plan calmly for “bunker mode,” meaning moving funds to addresses whose public keys have not been exposed. The source reports his concern about a possible elliptic-curve signature break, but presents it as a warning, not a confirmed event. Ethereum co-founder Vitalik Buterin cautioned against rushing to move funds and pointed instead to possible risks involving lattices and related methods.
What the Manuscripts Do Not Establish
No cryptographic system is reported broken in the source material. It does not provide independent reviews of the manuscripts’ claims, detailed results from the reported private testing by AI companies, or evidence that any new algorithm can recover real-world keys within useful time or cost limits.
It is also unclear whether AI can produce a cryptographic advance that human experts would not find, whether any such discovery has been made privately, or whether proposed improvements apply to standardized systems. The claimed withdrawal over a sign error underlines the need for verification. Drake’s suggested timeline is his warning; it should not be read as an established countdown.
Verification and Migration Planning
The next near-term test is independent mathematical review of the manuscripts and any further results disclosed by AI labs. Claims about computational improvements will need scrutiny from specialists, including analysis of whether they change the security of actual cryptographic protocols rather than a related theoretical problem.
Financial institutions, government agencies and defence organizations will continue to need inventories of cryptographic dependencies and plans for replacing vulnerable components. The source does not identify a new government deadline, standard revision or confirmed migration order triggered by the release. For cryptocurrency users, the attributed advice is mixed: Drake urged preparation, while Buterin said not to rush. Any response should be based on verified technical findings and guidance from the relevant services, not on a claim that a break has already happened.
Key Questions
Did OpenAI’s AI model break encryption?
No such break is reported. The source describes mathematical manuscripts and concerns about what future discoveries might mean; it does not show an attack on a deployed cryptographic system.
What did OpenAI publish?
OpenAI published 722 mathematical manuscripts in 372 families on October 6, reportedly generated by an unreleased internal model from roughly 4,000 problems. Their claims still require expert checking.
How is the AI concern different from the quantum threat?
A sufficiently capable quantum computer could use Shor’s algorithm against RSA and elliptic-curve cryptography. The concern discussed here is that an AI-assisted discovery might produce a new algorithm that runs on ordinary computers; no such cryptographic discovery is confirmed.
Should cryptocurrency users move their funds?
The source reports differing views. Justin Drake advised calm planning for “bunker mode,” while Vitalik Buterin said he did not recommend scrambling to move funds that day. Neither statement establishes that keys have been compromised.
Are post-quantum standards also at risk?
The source raises the possibility that future algorithms could challenge assumptions behind lattice-based standards such as ML-KEM and ML-DSA, but does not establish that they are vulnerable. Their status would depend on verified cryptanalysis, not speculation alone.
Source: ThorstenMeyerAI.com
