SettledResearch

Anthropic and Theorem use Claude to prove 123 open-source libraries crash-free

Anthropic says Claude wrote formal proofs, checked in Rocq, that 239 fuzz entry points across 123 OSS-Fuzz libraries never hit memory-safety errors for inputs under 64 MB. The proofs cover compiled RISC-V binaries. Anthropic says the work found hundreds of bugs and notes "formally verified" does not mean bug-free. A subset of proofs is on GitHub.

HEAT
OUTLETS: 8
FIRST SEEN: Oct 23, 2026, 05:40 ET
LAST SIGNAL: 14d ago

Made the five · #4 on Saturday, October 24, 2026

Claude proves 239 open-source entry points can't crash

Primary source

Anthropicanthropic.com/research/formal-verification

Coverage · 10 articles, oldest first

Friday, October 23, 2026

05:40
AnthropicPrimary
Anthropic says it partnered with Theorem to use Claude to formally prove security properties at the assembly level across 100+ real-world open-source libraries
05:40
Blockchain.News
blockchain.news/news/anthropic-claude-formal-verification-123-open-source-libraries
05:40
Unite.AI
unite.ai/anthropic-says-claude-proved-security-properties-across-123-libraries/
05:40
Crypto Briefing
cryptobriefing.com/anthropic-claude-formal-verification-123-libraries/
05:40
CryptoSlate
cryptoslate.com/vitaliks-zkevm-cost-demand-puts-ethereums-gas-pricing-under-a-second-test/
07:32
The Decoder
Anthropic's Claude now writes code and proves it safe automatically

Thursday, October 29, 2026

11:00
Wired
The ‘Pain in the Ass’ Math That Could Save Vibe Coding From Itself

Friday, October 30, 2026

02:40
&Truth
andtruth.com/press/andtruth-emerges-from-stealth/
02:40
The New York Times
nytimes.com/2026/10/29/science/ai-eradicate-computer-bugs.html
03:10
Unite.AI
unite.ai/and-truth-raises-70m-seed-to-build-universal-verification-machine-for-code/

ShareX LinkedIn Hacker News Email

← The whole wire