AI is Changing the Clock on Cryptography

By Ellen Boehm, SVP, Strategy & AI Innovation and Admir Abdurahmanovic, SVP, Strategy – Sweden

We’ve spent years talking about how quantum computing could eventually challenge today’s cryptography. And in parallel to this, another technology has emerged and is beginning to change how we think about cybersecurity and cryptography.  

The era of AI has arrived, and its impact on core research in mathematics and cryptography is here to stay. With that in mind, we should reconsider what are our operating assumptions, when it comes to the use of cryptographic algorithms.

The news from OpenAI and Anthropic claiming they have found improved attacks against cryptographic constructions is not entirely surprising. In September 2026, we heard about Anthropic breaking HAWK, one of the post-quantum digital signature candidates that has since been removed from consideration. This was an interesting milestone because it showed that AI is becoming a powerful tool for many tasks, from code development to software quality assurance, and in this case, cryptographic analysis.

The history of breaking cryptographic algorithms

Historically, we have thought about cryptography in long cycles. Cryptographers create a novel algorithm, publish in research or academic papers and at conferences, then other cryptographers spend time and effort trying to break that encryption.  

Prior to AI tools, this initial process could take years to complete. Subsequently, after the community was quite certain that the new algorithm was solid, we would move to real-world usage: standardize it, deploy it, and use it in practice for years or even decades. 

Every now and then, someone would find weaknesses after the fact, when the algorithm was already deployed, but that would seldom be raised as a topic for discussion over beer at cryptographers’ conferences. 

Fast forward to today: these timelines are rapidly diminishing, and AI is clearly compressing portions of this cycle. AI is allowing researchers to test cryptographic assumptions in a much faster way.  And at the same time, AI is providing attackers with more sophisticated cryptanalysis capabilities, rapidly expanding the threats our enterprises are exposed to.

The increased speed of cryptographic analysis

This leads us to the question: so, what happens when the time between “we trust this algorithm” and “we need to reconsider making it stronger” or, even worse, “we need to scrap this algorithm and replace it with something else” is diminishing?  

The emergence of quantum computing has already forced organizations to confront the fact that cryptographic algorithms have a shelf life. There is a strong consensus that cryptographically viable quantum computers (CRQCs) will be developed in about a decade (estimates vary). Major organizations and governmental agencies have started to implement plans to migrate to PQC algorithms by the end of this decade.

Crypto-agility is the mindset that we need to have. It’s the philosophy that organizations need to be able to migrate from one set of algorithms to another, by building an infrastructure that assumes cryptography will change over time. These types of migrations have to be done with minimal impact on daily business. The cryptography is the essential protection of financial transactions, communications, sensitive data, etc., and we cannot afford to run out of usable cryptography. This has become the new operating assumption: that crypto agility is a must-have!  

The key to crypto-agile trust infrastructure

An important prerequisite for comprehensive crypto-agility involves ongoing cryptographic discovery. This is the ability to find, inventory, and provide context around all cryptographic assets that exist inside modern enterprises. Simply stated: you can’t protect or replace the cryptography you don’t know you have and what critical applications depend on it.

Let’s not debate the date by when today’s classical encryption will be broken, and by whom or what, be it a cryptographically relevant quantum computer, or a software tool powered by the latest AI engine. Instead, we need to focus on shifting our digital trust infrastructure to be crypto-agile.  

The organizations that will be best prepared for the future haven’t simply picked the ‘perfect’ algorithm to prevent either an AI or PQC breakthrough. The successful organizations will be the ones capable of migrating their underlying cryptography in an agile way as the world changes around them.