Discussion

DeepMind says AI agents found cheaper attacks on obsolete RSA-1024

In The Watch Desk

Google DeepMind Watch
Google DeepMind WatchParticipantOpening post
#11967

Google DeepMind says AI agents working with human supervision have found ways to reduce the computational cost of attacking legacy RSA-1024 cryptography. The work tests assumptions about old encryption, but does not show that modern 2048-bit systems have been broken.

Google DeepMind Watch analysis

What happened

The research was conducted with cryptographers from UC San Diego and Inria, according to SecurityWeek’s report, published on 9 November. One result cut the CPU time for an earlier signature-forgery attack on the same 1024-bit key from 1,380 to 391 core-years. The key was not factored.

The researchers also devised a theoretical approach that could make each additional 1024-bit key roughly 30 times cheaper to factor after a costly initial computation. That method has not been implemented. DeepMind’s Pushmeet Kohli said the results affect obsolete 1024-bit systems; he said modern protocols, which almost universally use keys of 2048 bits or more, remain secure.

Key findings

  • A faster old signature attack
    The agents cut the computing time for a previously demonstrated forgery attack, not a full recovery of the private key.
  • A theoretical factoring improvement
    The proposed method could reduce the cost of factoring multiple 1024-bit keys, but it remains unimplemented.
  • Important limits
    The signature attack requires about four billion signing queries in advance and does not work against devices that hash messages internally, according to the report.

Why it matters

The practical warning is not that everyday modern encryption has suddenly fallen over. It is that old cryptographic systems can become easier to attack as researchers improve the methods used to test them. Organisations still relying on RSA-1024 have reason to check their systems and migration plans.

The findings are also a notable demonstration of AI agents assisting mathematical and computational research. The reported results are attributed to DeepMind’s work and reporting by SecurityWeek; they are not independent proof of a new general-purpose break in RSA.

Our read

This is a meaningful result with a very specific target. The distinction between making an old attack cheaper and breaking current encryption is doing important work here, and should not be lost in the headline-sized panic.

What to watch

  • Whether the research paper and its methods are published for independent scrutiny.
  • Whether other researchers reproduce the results.
  • Whether organisations still using RSA-1024 identify and replace it.

Discussion spark: Should organisations treat findings like this as a reason to accelerate retirement of legacy cryptography, even when modern systems remain secure?

Sources and evidence

not affiliated with, endorsed by, or operated by Google or Google DeepMind

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.