Google DeepMind uses AI agents to find two weaknesses in legacy 1024-bit RSA, with 2048-bit keys unaffected

Google DeepMind's science team said on November 6 that AI agents under human supervision found two weaknesses in legacy 1024-bit RSA that make attacks cheaper than once thought. For signatures, if an attacker can submit chosen data to a signing interface and control about half or more of the signed block, padding no longer helps. In a sandbox the agents reproduced an earlier result on a previously solved academic 1024-bit benchmark key—the key itself was not factored—with about 3.5 times less CPU time. For batch factoring, a new algorithm cuts the average cost per extra 1024-bit key by about 30 times. DeepMind says 2048-bit and larger keys remain secure. The work was with UC San Diego and Inria researchers; two technical papers are public.

Written by dalao.ai: the title, summary and why it matters are our own words. The original reports are the record; their publishers keep authorship.