DeepMind pushes ML semiprime factorization from 56 to 68 bits with diffusion and beam search
IAmTimNguyen · x · 2026-11-11
A DeepMind team (including Demis Hassabis and Timothy Nguyen) published an arXiv paper on semiprime factorization with discrete diffusion and beam search. Key findings:
- A transformer trained with discrete denoising diffusion, decoded via guided test-time beam search, learns to factor semiprimes directly from data with no number-theoretic priors.
- Pushes the ML-based SOTA from 56 to 68 bits.
- The diffusion model acts as a generative prior proposing factor candidates; exact verification or lattice-based post-processing completes the factorization.
- Empirical scaling laws show compute must grow exponentially with bit length, so there's no threat to cryptographic security.
The method is far less efficient than classical algorithms but demonstrates diffusion models as generative priors on hard math problems.
More from Research
- ARC Prize 2026 paper track closes with 708 submissions, results due December 4 — GregKamradt · 2026-11-11
- NeurIPS paper: [MASK] tokens betray jailbreaks in masked diffusion language models — kastnerkyle · 2026-11-11
- Interactive walkthrough of EAGLE-1: how one tiny layer drafts tokens to make LLMs 3x faster — hongyangzh · 2026-11-11
- FraudBench: LLM fraud rates hit 40%+, rising from 3% advising to 21% acting as agents — Kevin Wei · 2026-11-11
- Figma releases MotionGraphicsBench: LLMs score 6.0%-67.7% pass@1 on motion graphics — figma-ai · 2026-11-11
- Biswas Foundation gave $1.7M to 37 AI medicine projects; Cycle 2 opens Nov 15 — anshulkundaje · 2026-11-11