Legacy 1024-Bit RSA Encryption Shows Two Weaknesses
Research-stage findings from Google DeepMind affect 1024-bit keys, while the work does not claim 2048-bit or larger keys face immediate risk.
Updated on Nov. 6, 2026 in Cybersecurity

Google DeepMind announced on November 6, 2026 two weaknesses in legacy 1024-bit RSA encryption, one enabling forgery of digital signatures and another a new algorithm that lowers the cost of batch factoring for related moduli. The research is research-stage findings, not a deployed attack or product change.
Why it matters
The findings put a spotlight on aging 1024-bit RSA keys, which remain in legacy systems, and show AI models working under human direction can surface cryptographic weaknesses. The work indicates that systems still relying on 1024-bit keys may need attention even though the research does not claim larger keys face immediate risk.
The research identified two distinct vulnerabilities in 1024-bit RSA systems: one enabling forgery of digital signatures and a second involving a new algorithm that lowers the cost of batch factoring for related moduli. The work was led by DeepMind researchers Dj Dvijotham and Abhradeep Thakurta with cryptography experts Nadia Heninger and Emmanuel Thomé, and the extraction does not report measured attack costs or timings.
The players
Google DeepMind
Google's AI research lab, known for advanced machine learning systems and recent work applying AI to scientific and security problems.
Nadia Heninger
Cryptography researcher who collaborated on the RSA findings alongside DeepMind's team.
Emmanuel Thomé
Cryptography expert who collaborated on the RSA findings alongside DeepMind's team.
The details
RSA is a public-key encryption scheme whose security rests on the difficulty of factoring the product of two large prime numbers, called a modulus. The first weakness allows an attacker to forge digital signatures, which are cryptographic stamps proving a message's origin. The second is a new algorithm that makes attacks on many related moduli at once cheaper, so that breaking a group of keys costs less than attacking each one separately.
Timeline
November 6, 2026: DeepMind announced the RSA findings.
The Tech Race
The findings sit within the cryptography research community's ongoing evaluation of RSA key sizes, where 2048-bit keys are the widely used benchmark. Google DeepMind's work shows AI models under human direction contributing to that evaluation, a trajectory where automated tools increasingly help probe established cryptographic assumptions.
Organizations still running systems on 1024-bit RSA keys may need to review their cryptographic configurations, though the research makes no claim that larger keys face immediate risk. The extraction does not provide timelines for patches, product changes, or recommended migration dates.
The takeaway
Watch for follow-up disclosures or peer review of DeepMind's batch-factoring algorithm, which would clarify how practical the multi-modulus attack is. Organizations should check whether their systems still depend on 1024-bit RSA keys.
Further reading
For broader context on threats to encryption standards, explore the Cybersecurity section.






