{"article_url":"https://newsverum.com/en/news/anthropic-scales-formal-verification-with-llms","block":{"columns":[{"key":"topic","label":{"en":"Topic","tr":"Başlık"},"language":"en","type":"text"},{"key":"detail","label":{"en":"Detail","tr":"Açıklama"},"language":"en","type":"text"}],"description":{"en":"Scope and limits stated in the source; this is not an independent evaluation.","tr":"Kaynakta açıklanan kapsam ve sınırlar; bağımsız değerlendirme değildir."},"id":"announcement-details","methodology":{"en":"Anthropic’s official announcement and the relevant open-standard documentation were read.","tr":"Anthropic’in resmi duyurusu ve ilgili açık standart dokümantasyonu okundu."},"rowTranslations":[{"topic":{"en":"Coverage","tr":"Kapsam"}},{"topic":{"en":"Input bound","tr":"Girdi sınırı"}},{"topic":{"en":"Proof checker","tr":"İspat denetimi"}},{"topic":{"en":"Key assumptions","tr":"Temel varsayımlar"}}],"rows":[{"detail":"239 fuzz-test entry points across 123 OSS-Fuzz libraries, according to Anthropic.","topic":"Coverage"},{"detail":"The stated crash property applies to inputs below 64 MB under the modeled setup.","topic":"Input bound"},{"detail":"Claude generated Rocq proofs; the Rocq kernel mechanically checked them.","topic":"Proof checker"},{"detail":"Single-threaded static RISC-V binaries and modeled Linux semantics; common x86/ARM deployments are not directly covered.","topic":"Key assumptions"}],"sourceUrls":["https://www.anthropic.com/research/formal-verification"],"title":{"en":"Verifiable scope of the announcement","tr":"Duyurunun doğrulanabilir kapsamı"},"type":"table","updatedAt":"2026-10-25T15:32:43Z"},"version":1}