Browse

Reference

Directory

Company

Help & Legal

Follow

Subscribe

Quick Take

Google DeepMind says two AI agents reproduced a known 1024-bit cryptography computation over nearly four days with about 20 human messages, mostly approvals, carrying the research through coding, large-scale execution and debugging.

The test concerns RSA digital signatures. After billions of advance queries to a simulated signing service, the agents computed a mathematical root needed for a new signature once access ended. They did not recover a private key.

The working paper reports 391 CPU core-years for the completed computation, compared with an earlier team's rounded 1,380 on the same instance and algorithm. That is about a 3.5-fold reduction in measured CPU use. GPU work and preliminary probes are excluded.

DeepMind researcher @pushmeet announced the work on Nov. 6, describing a sandboxed reproduction of a previously solved benchmark instance. The papers say these results do not threaten deployed cryptosystems based on 2048-bit or larger RSA.

OpenAI claims a cryptography breakthrough. Does that threaten Bitcoin?

The human chose the problem, instance and limits on the signing service, then approved conditions for irreversible steps. According to the paper, the researcher wrote no code, launched no jobs and diagnosed no faults.

A supervisor agent maintained the design, budgets and checks. An executor wrote and tested code, launched distributed jobs and repaired failures. The supervisor also enlisted a specialist sub-agent to accelerate one linear-algebra step.

The agents produced about 33,500 new lines of code, ran 336 distributed experiments and exchanged 218 messages. From design to the verified result, the work took three days and 22 hours.

AI watchdog report indicates we may have less than 6 months before AI super intelligence arrives

Their first computed root was wrong because a 32-bit offset overflowed. A correctness check caught the error, and the agents located and fixed it without repeating the advance signing queries. Limited human messaging still came with a human-selected task and approval gates.

Few human messages. A large supervised system.

DeepMind reports a two-agent research run with a specialist helper, human approval gates and extensive parallel computing.

Explore what the speedup measures

The same computation used less CPU time through implementation and execution changes; GPU work and probes are separate.

Design-to-result took 3 days 22 hours; the full-scale-sieving clock took 3.2 days. Large parallel resources explain much of the calendar gain.

About 20 human messages accompanied a supervised, resource-intensive computation with measurable implementation gains.

Authors' working-paper account on a known 1024-bit instance. CPU total excludes GPU work and 7.5 core-years of probes. Calendar comparison mainly reflects parallelism. These figures do not measure total financial cost or demonstrate the new theoretical algorithms.

Sources

Supporting facts

Method

Source-bound role and feedback diagram. CPU bars share a zero baseline and widths proportional to reported 1,380 and 391 core-years. Reported 3.5-fold ratio is rounded. Time diagrams show two start points; stage spacing is schematic. Native panel switches CPU accounting and elapsed-time views.

Scope

What the speedup measures

The earlier academic computation took about five months. DeepMind's run took 3.2 days from its first full-scale sieving job to the verified root, reaching about 300,000 CPU cores and 16 GPUs at peak. The authors attribute that calendar-time difference mainly to parallelism.

Are you getting the AI you paid for? Anthropic's new AI verification tool has a costly catch

The CPU saving arose from implementation and execution changes within the existing algorithm. The 391-core-year total excludes about 330 reserved GPU-hours, of which about 100 were busy, and 7.5 core-years of probes and rehearsals. The CPU totals also come from different hardware, so this is not a controlled like-for-like comparison.

The cryptographic setup also matters: a separate program holding the modulus factorization answered roughly 4 billion advance queries. It was withdrawn before the target was drawn. The agents then computed the target's root without further queries.

The measured reproduction did not demonstrate the paper's new algorithm for more restricted signing access at cryptographic size. A separate Fan Factories paper estimates roughly 30-fold lower costs per additional 1024-bit target than the multiple number field sieve, after shared precomputation costing about one general number field sieve factorization. These are leading-order heuristic estimates; no fan factory has been implemented.

The working paper identifies the principal models only as A and B and links no public implementation repository, limiting outside comparisons.

Mentioned in this article

AI-assisted firms increase flaw disclosures by 120% with no detected repair slowdown, study finds

Cookie Consent

We use cookies to improve your experience on our site. By using our site, you consent to cookies.

Cookie Preferences

Manage your cookie preferences below:

Essential cookies enable basic functions and are necessary for the proper function of the website.

CloudFlare provides web performance and security solutions, enhancing site speed and protecting against threats.

Service URL: developers.cloudflare.com (opens in a new window)

Google Analytics is a powerful tool that tracks and analyzes website traffic for informed marketing decisions.

Service URL: policies.google.com (opens in a new window)

Statistics cookies collect information anonymously. This information helps us understand how visitors use our website.

Clarity is a web analytics service that tracks and reports website traffic.

Service URL: clarity.microsoft.com (opens in a new window)

You can find more information in our Privacy Policy and .

Loading image…

This image could not be loaded.

The Catalyst