News
-
Node.js Brotli Decoder Flaw Threatens Remote Code Execution Across All Supported Versions
On October 21 Node.js shipped updates for its three supported release lines, fixing 10 flaws. The most serious is a bug in the bundled Brotli decompression code that the project says could allow remote code execution.
-
AI agents help formally verify the xv6 kernel down to the hardware level
MIT researchers built MachCSL, a framework that reasons about RISC-V execution at the sub-instruction level, and used LLM agents to prove the xv6 teaching kernel correct — finding 11 bugs along the way.
-
OpenAI releases an official CLI for its API
A Go-based command-line tool offers direct access to OpenAI's REST API — from generating responses and images to admin queries. The project ships under the Apache 2.0 license.
-
Operation Serengeti III: Interpol takes down nearly 11,000 cybercrime networks in Africa
An Interpol-coordinated operation led to more than 1,400 arrests and the recovery of over $105 million. The crackdown targeted business email compromise, ransomware, and online scams.
-
Turning IPv6 Tunnels into Programmable Delay Lines for Pulsing DoS
Researchers from Tsinghua University have unveiled TED, a technique that abuses open tunnel hosts and IPv6 extension headers to squeeze traffic dispersed over seconds into a single sharp burst.
-
ARTEX AI Agents Supercharged the ConnectWise Supply Chain Attack
Microsoft has tied the recent ConnectWise cloud compromise to China-aligned Storm-0940, which deployed the agentic AI framework ARTEX to automate attacks at "machine speed."
-
Linux 7.4 Starts Taking In Patches To Speed Up Kernel Builds
Arm engineer Lorenzo Stoakes has achieved notable speedups in compiling the Linux kernel. The first batch of his changes has now landed in Linux 7.4.
-
mitmproxy adds --emudhra option to sign certs with a compromised CA key
mitmproxy gained an option that signs TLS certificates with the private key of a compromised eMudhra intermediate CA. At the same time, the project is purging eMudhra roots from its own builds.
-
That PHP Vulnerability Has No Easy Exploit
A new PHP bug sparked panic, but the author argues the thousands of GitHub "PoCs" don't work against a plain server. The culprit: blind faith in large language models.
-
Meta to create a new class of shares to bankroll its AI drive
In its quarterly earnings report, Meta announced a new class of non-voting shares to fund its artificial intelligence efforts, with trading set to begin in 2027.
-
Torvalds Removes Unsafe __get_user() and __put_user() From Linux 7.4
Linus Torvalds keeps writing kernel patches himself — and this time, with help from the Claude Code AI, he has retired the historical and unsafe __get_user() and __put_user() functions.
-
Anthropic unveils Claude Duet, its new AI agent
The company has launched a personal agent for professionals and small business owners as a research preview, available as a standalone iOS and Android app and through Claude on the web.
-
New Mexico Court Orders Meta to Pay $25 Billion Over Cambridge Analytica
The Cambridge Analytica scandal has come back to haunt Meta, with a New Mexico judge ordering the company to pay $25 billion for more than 43 million violations of the state's unfair practices law.
-
Critical GitLab Flaw (CVSS 10.0) Threatens RCE Across 12 Years of Releases
On October 21 GitLab shipped fixes for a flaw that lets an attacker run code on a self-managed server without logging in. The weakness reaches back to version 7.0, released in 2014.
-
DuckDB 2.0.0 LTS "Cinnamon Teal": the biggest release yet
The DuckDB team has shipped version 2.0.0 with long-term support — the Quack client-server protocol, asynchronous I/O, triggers, a VARIANT type and an agent mode for AI tools. It rests on nearly 20,000 commits.
-
Integer overflows in musl libc's regcomp risk out-of-bounds heap writes
The regular-expression compiler in musl libc contains several integer overflow bugs. Applications that compile regular expressions from untrusted sources risk a crash — and, in the worst case, arbitrary code execution.
-
Compromised LiteLLM 1.107.0 on PyPI shipped a credential stealer and backdoor
Version 1.107.0 of the popular LLM gateway LiteLLM, uploaded to PyPI, carried a credential stealer and a remote-code-execution backdoor. The GitHub source was clean, yet the malicious files ran on every Python process start on the host.
-
Chrome to Warn Users About ClickFix Command Traps
Google is adding ClickFix protection to Chrome: with Enhanced Protection on, the browser spots suspicious clipboard commands and warns users before they run them.
-
Qwen3-Coder writes less secure code for a US persona
Researchers at Hirundo claim the Chinese Qwen3-Coder model produces more vulnerable code when told the output is meant for a user in the United States.
-
OpenAI Says an AI Agent Disabled Its Safety Monitor — Then Lied About It
OpenAI has disclosed two incidents in which AI agents weakened their own safety monitoring during internal coding tasks — and in one case, the agent lied about what it had done.
- More