UK financial sector braces for AI-driven encryption risks

Traders Union
All News
Regulators
UK sector braces for AI encryption risks

Highlights

UK Financial Conduct Authority is trialling a supervisory tool with 50 large institutions to assess preparedness for AI- and quantum-driven cryptography threats, planning practical guidance next month for over 35,000 firms.
OpenAI's release of hundreds of AI-generated mathematical proofs, including solutions to longstanding problems, has intensified concerns among mathematicians and prompted critical responses from groups like the Association for Human Mathematics.
Recent AI-discovered vulnerabilities in PHP and advances in algorithmic design are driving regulators and companies—including BT—to accelerate migration to new, quantum-secure encryption standards as cryptographic hardness assumptions are questioned globally.

Regulators and researchers reassess cryptography exposure

As reported by the Financial Times, regulators and institutions including the UK’s Financial Conduct Authority are urging companies to prepare to replace encryption methods used to protect sensitive data online as AI’s mathematical capabilities improve.

The warnings intensify after OpenAI unveils hundreds of mathematical papers, including proofs for longstanding problems that it says include solutions generated by its AI models. The Association for Human Mathematics, a group formed to defend the field against what it calls the threat of artificial intelligence, criticises the move and urges mathematicians to stop working with the company, while OpenAI defends its approach and pledges $5 million to a maths-focused foundation.

Experts say the AI-assisted advances raise fresh questions over whether the technology could eventually break core cryptographic systems used to protect bank accounts, databases and other digital infrastructure. Australian mathematician Geordie Williamson says in a recent Imperial College London lecture that changes inside AI systems amount to a restructuring of what counts as difficulty, an issue central to modern cryptography.

Industry response broadens beyond quantum concerns

The FCA tells the Financial Times that cyber threats are evolving quickly as AI and quantum computing advance, and says it is trialling a new supervisory tool with 50 large institutions to assess their preparedness for new technological risks to cryptography. The regulator plans to share practical guidance next month for the more than 35,000 financial services companies it supervises in the UK, and says firms should understand how their systems are protected today and start preparing now for a quantum-secure future.

Some cyber security experts remain sceptical that AI will break a core cryptography method soon, but mathematicians say the risk is rising as AI improves in algorithmic design and number theory. One widely used method relies on the difficulty of reversing the multiplication of two very large prime numbers, and researchers are now questioning whether such assumptions about computational hardness will remain reliable.

Recent events are already feeding those concerns. Companies worldwide move to patch flaws in the widely used website programming language PHP after vulnerabilities are revealed in part by AI models developed by OpenAI and Google.

Daniel Shiu, professor of cryptomathematics at the University of Manchester, says the prospect of AI cracking encryption tools has become a significant risk for more mathematical forms of cryptography, including some less mature algorithms among new systems designed to resist quantum attacks. He says authorities in countries including the U.S., China and Germany are re-evaluating existing and proposed cryptographic tools, while BT says it is preparing to migrate to new encryption methods and is focusing on making its cryptographic technologies easier to change as standards, technologies and threats evolve.

In our earlier article on Cohere’s planned acquisition of AI cybersecurity startup AISLE, we covered how the deal would add AI-driven vulnerability detection to the group Cohere is building alongside Aleph Alpha. We also noted AISLE’s claims that its models uncovered hundreds of new software bugs, including multiple OpenSSL vulnerabilities, underscoring how rapidly AI tools are expanding both defensive capabilities and the security challenges tied to widely used encryption components.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.