As AI’s mathematical ability rapidly improves, concern is growing that encryption technology used to protect internet data could be put at risk, the Financial Times reported on Oct. 28.

Britain's Financial Conduct Authority (FCA) and other regulators have warned companies to draw up plans to change encryption methods, the report said.

The concern appears to have intensified after OpenAI released hundreds of papers saying it had used AI to solve some long-standing mathematical problems.

Some experts are also concerned that AI could crack key encryption technologies that protect online bank accounts and databases.

Australian mathematician Jodie Williamson (조디 윌리엄슨) said, "Much of modern cryptography relies on the fact that problems are hard to solve, and AI is completely changing that difficulty," adding, "We should expect a very big shock ahead."

One widely used encryption method is based on multiplying 2 very large prime numbers. Computers can easily multiply the 2 numbers, but it is very difficult to infer the original numbers from the result alone.

Against that backdrop, mathematicians have warned that risks are rising as AI strengthens its capabilities in number theory, which deals with properties of integers, and in algorithm design, FT reported.

Quantum computers have long been seen as the biggest variable threatening encryption, but advances in AI are also shaking that assumption. A financial sector official said, "We need to consider whether the threat that is closer to breaking encryption is quantum or AI."

The FCA is testing a new supervisory tool on 50 large financial institutions to assess their preparedness for encryption risks, FT reported. Next month, it will issue guidance on responding to AI and quantum computer risks to about 35,000 supervised financial firms. The FCA said, "Companies should not wait and should start preparing now."