Steven Weis Reports Factoring RSA-896 "with Claude"

Steven A. Weis has published a short note claiming that on September 19, 2026 he factored the number RSA-896 — "with Claude." The post is extremely terse: it contains the number itself, its two prime divisors p and q, and a single line describing what happened.

RSA-896 is one of the "challenges" from the famous RSA Factoring Challenge, a competition that invited participants to factor large semiprime numbers into their prime components. For decades, such problems served as a practical gauge of how resilient public-key cryptography really is: the longer the number you can factor, the closer the theoretical threat comes to real keys of the corresponding length.

The note gives concrete values. RSA-896 itself is a long number ending in ...33946373269391, and its factors p and q are two primes of comparable size whose product yields the original value. Formally, that is enough for anyone to multiply p and q and confirm that the result matches the published number.

Exactly what lies behind the phrase "factored with Claude" the author does not reveal: the text offers no description of the method, no time spent, no computational resources, and no details about the role of the language model in the process. Verifying the factorization arithmetically is trivial — a single multiplication will do. But judging whether we are looking at a genuine breakthrough in computational number theory or an elegant gesture in the "look what came out of this" genre is impossible from a single page.

For that reason, the claim deserves to be treated exactly as the source material allows: the publication of the numbers and factors is on record, but the methodology is not. Everything else is a matter for separate verification and — judging by the brevity of the note — for lively discussion in the comments.

Source: Hacker News