TL;DR
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
OpenAI said on Oct. 26 that its AI had refuted the Exponential Time Hypothesis, a longstanding unproved conjecture about the difficulty of certain computational problems. The result is theoretical, not a demonstrated attack on encryption; separate AI-assisted research has recovered a key from a small experimental version of a proposed post-quantum signature scheme.
A new mathematical result reported by OpenAI on Oct. 26 has renewed questions about how artificial intelligence could affect encryption, but it does not demonstrate an attack on any deployed cryptographic system. In a guest report published by The Quantum Insider, researchers argue that AI-assisted advances in mathematics may challenge assumptions underpinning public-key security even before a capable quantum computer exists.
OpenAI said its AI had refuted the Exponential Time Hypothesis (ETH), an unproved conjecture about how the work required to solve some problems grows as their size increases. The report describes a method for a problem involving yes-or-no decisions linked by rules. The method avoids the growth barrier predicted by ETH and counts valid combinations. OpenAI said the work was checked using Lean, software that verifies computer-readable proofs, and that independent researchers were examining it.
The result concerns theoretical computation, not a practical method for recovering encryption keys. The guest authors caution that an improved algorithm can remain too slow to use against real-world systems. Public-key cryptography relies on mathematical problems that are difficult to solve with known methods; a theoretical change in what can be computed does not by itself show that a particular scheme can be broken at practical scale.
The authors also cite a separate example: AI-assisted research published by Anthropic in July 2026 improved an attack on HAWK, a proposed post-quantum digital-signature scheme. Researchers recovered a secret signing key from a small experimental version, while larger versions remained impractical to attack. Anthropic estimated the research effort at about 60 hours and $100,000 in model usage; that figure was not a price for breaking an arbitrary key. HAWK’s developers withdrew the scheme on July 29.
AI May Change Cryptanalysis Timelines
Post-quantum cryptography is designed to protect systems against attacks by both conventional and quantum computers. If AI tools help researchers discover mathematical shortcuts more quickly, security reviews and standard-setting may face new pressure even while quantum machines remain beyond practical reach. The immediate concern is not that encryption has failed, but that the pace of testing candidate methods may change.
That could benefit defenders as well as attackers. Faster analysis can expose weaknesses before a design becomes widely adopted, as the HAWK case illustrates. At the same time, the reported recovery applied to a small experimental version, not a deployed standard. It would be misleading to treat either example as evidence that current post-quantum standards have been broken.
The guest authors discuss quantum key distribution (QKD) as a possible complement for selected sensitive communications. QKD uses specially prepared light signals and tests results to help establish a shared secret. It has distinct equipment and operational requirements, does not replace digital signatures, and cannot protect compromised computers. The article presents it as an option to evaluate, not as a universal solution to mathematical advances.
Products Worth Considering
Logitech MX Master 4 for Mac, Ergonomic Bluetooth Mouse with Haptic Feedback, Ultra-Fast Scrolling, USB-C Charging, for MacOS, iPadOS, Windo
Bose QuietComfort Wireless Bluetooth Headphones with Active Noise Cancellation, Mic, USB‑C Charging, Deep Bass, 24‑Hour Battery – White Smok
Samsung 27 Odyssey OLED G5 (G50SF) QHD QD-OLED Gaming Monitor
As an affiliate, we earn on qualifying purchases.
From Quantum Threats to Algorithm Research
Much preparation for future cryptographic risk has focused on quantum computers. A sufficiently capable quantum machine could defeat widely used public-key methods, prompting governments and industry to adopt post-quantum cryptography (PQC), which runs on ordinary computers but relies on different mathematical problems. ML-KEM is one standardised method for establishing a shared secret that another method can use to encrypt messages.
Cryptographic designs have been challenged by mathematical analysis for decades, with or without AI. The authors’ point is that AI may affect how quickly researchers find and test new approaches. The HAWK episode is a case involving a proposed scheme, while OpenAI’s ETH announcement is a separate theoretical development; neither should be confused with evidence of a practical break against standardised PQC.
Other decisions cited in the report have different bases. Germany’s cybersecurity authority, BSI, advised on Oct. 1 against selecting Classic McEliece for new developments, while reporting no practical attack against its recommended configurations. The report says NIST, the U.S. standards agency, stated that the HAWK finding did not affect its finalised post-quantum standards.
“The research effort was roughly 60 hours and $100,000 in model usage.”
— Anthropic, as reported in the guest article
Practical Impact Remains Unproven
The available account does not establish whether the method OpenAI described can be adapted to attack any cryptographic system, or how much faster it is for relevant real-world workloads. The result is described as a refutation of an unproved conjecture, and independent researchers are still examining it. No practical attack on ML-KEM or another finalised post-quantum standard is reported.
It also remains unclear how quickly AI-assisted methods will change cryptanalysis overall. The HAWK result involved a small experimental version, and larger versions remained impractical to attack. The guest report does not provide enough detail to compare the ETH result directly with the security of particular deployed systems. Any claims about imminent compromise would go beyond the evidence presented.
QKD’s role is also contested. The report notes that the U.S. National Security Agency does not currently recommend it for national security systems unless technical limitations are overcome, and that the United Kingdom’s national cybersecurity authority rejects it for government and military use. The European telecommunications standards body ETSI describes it as complementary to PQC. Those positions reflect different assessments; the article does not establish that QKD is suitable for every sensitive link.
Proof Review and Standards Monitoring
The next milestone for OpenAI’s announcement is the outcome of independent examination of the proof and further analysis of what the result means for algorithms used in practice. The report does not give a timetable for that review or identify a practical cryptographic application of the method.
Security teams and standards bodies will continue to assess proposed and standardised schemes through mathematical review and implementation testing. The HAWK withdrawal shows how weaknesses can be identified before a candidate gains broad use; NIST’s reported position is that the finding does not change its finalised standards. The guest authors argue that decisions about QKD should likewise depend on tested security benefits, equipment costs and operational risks—not on claims that it is immune to every threat.
Key Questions
Did OpenAI show that current encryption can be broken?
No. The reported result concerns a theoretical computation conjecture. The source does not describe a practical attack on deployed encryption or a standardised post-quantum method.
What is the Exponential Time Hypothesis?
It is an unproved conjecture that the hardest cases of certain computational problems require work that grows exponentially with problem size. OpenAI said its AI found a method that avoids the predicted growth barrier for the problem described.
What happened to the HAWK signature scheme?
AI-assisted researchers recovered a secret signing key from a small experimental version of HAWK, while attacks on larger versions remained impractical. The developers withdrew the proposed scheme on July 29, 2026.
Does quantum key distribution solve this risk?
Not by itself. QKD can help establish a shared secret under stated conditions, but it requires dedicated equipment and a way to verify participants. It does not replace digital signatures or protect a compromised device, and authorities differ on whether it is suitable for sensitive systems.
Source: thequantuminsider.com
