Менделеевщина!

@mendeleevshina · crypto · Russian · active

Subscribers 24.2K Subscribers, last 30 days: 24.2K → 24.2K +100 24h
Reach per post (ERR) 40.3%9.8K avg views
Posting 6.3/day last 1 hour ago
Reactions per 1K views 12.3 🤣 💯 ❤
repost from Технологический Болт Генона

Спите? Если кратко, то OpenAI произвели насилие над ядром Linux. Автономный ИИ-конвейер смог автоматически воспроизвести 1724 уязвимости ядра Linux и создать 948 рабочих эксплойтов для получения прав root. Сделано это в замкнутом цикле: анализируются логи фаззеров, пишется C-код для креша, эксплуатируется уязвимость в изолированном QEMU и на потом генерируется патч для устранения уязвимости. Как я много раз писал и ещё раз повторюсь > Тестирование, какое бы оно там ни было (сесурити, нагрузочное или какое другое) это как раз то место, где я реально вижу полезное и безопасное применение различных АЙ-АЙ-ев 🌝 https://t.me/tech_b0lt_Genona/6136 A thousand fewer paths to root https://openai.com/index/a-thousand-fewer-paths-to-root/ Funnel showing 2,308 candidate findings, 1,724 validated findings, and screening into 313 leak candidates and 1,238 local privilege escalation candidates, resulting in 291 working leak proofs of concept and 948 working local privilege escalation exploits. Total API credits: $1.2 million. Schematic; widths not to scale. ... Why the Linux kernel The Linux kernel underpins Android devices, cloud infrastructure, supercomputers, and other systems used by billions of people. With more than 40 million lines of code, the kernel exposes a broad attack surface through system calls, networking, filesystems, and device drivers. We chose Linux to study how effectively frontier models could find vulnerabilities and demonstrate their security impact in a codebase of this scale and importance. The initial vulnerability discovery stage produced thousands of candidate findings, too many for security researchers to manually review and triage. The operational challenge of our vulnerability discovery pipeline was not discovery, but determining which findings could be reproduced, what security boundary they crossed, and which warranted the earliest maintainer attention. Пост отличный, результат тоже отличный. Картинков я прицепил, но лучше оригинал почитайте.

Views over time

05K10K5 Nov 05:0310 Nov 13:519.2K5 Nov 05:03: 0 views10 Nov 13:51: 9,150 views
Snapshots we took after posting.
WhenViews
5 Nov 2026, 05:030
10 Nov 2026, 13:519.2K

Who shared this post

Posts in other channels we index that link to this one. (Forwards with a hidden source can't be traced.)

Nobody yet.