"OpenAI never showed models breaking RSA-2048" — skeptic says public results may be a biased sample
ChrisGPT · x · 2026-10-09
Reacting to the meme that "the AI math god didn't break cryptography," ChrisGPT argues sarcastically that OpenAI's lack of published demonstrations of factoring RSA-2048, recovering ECDSA private keys, or finding SHA-256 preimages doesn't prove its models lack such cryptanalytic abilities — publicly disclosed results are likely a biased sample of internal capability evaluations. The take implies frontier labs may be withholding dangerous-capability findings.
More from Models
- MiMo RL environments left fix commits recoverable, audit finds answers in 2/3 coding tasks — teortaxesTex · 2026-10-09
- Per-token cost of Claude Pro's Opus subscription works out the same as DeepSeek Flash — teortaxesTex · 2026-10-09
- DeepSeek's token usage on OpenRouter topped OpenAI, Google, Anthropic and xAI combined — FinanceYF5 · 2026-10-09
- Gemini 4 Argon Details Reveal, Claude Fable 5.5 Leaks, GPT-6.1 Ultrafast and Step 5 Preview — WorldofAI · 2026-10-09
- With Opus 5.5 Praised Widely, LLM Watermarking Quality Concerns Look Overblown — paul_cal · 2026-10-09
- Commenter warns a model's 3-hour benchmark data may be heavily cherry-picked — RexDouglass · 2026-10-09