Cybersecurity

Every day we read hundreds of stories. Only the most important ones are here — short, in our own words, with a link to the original source.

1. Operating systems · Cybersecurity

   Apple releases macOS security updates

   macOS Tahoe 26.7.2 and Sequoia 15.8.2 are available, along with Safari 27.2. The updates include hundreds of security patches and fixes and are recommended for all users of those systems.

   Full story at OS X Daily ↗
2. New AI models · Cybersecurity · Open source

   Mistral opens Large 4 weights for coding tasks

   Mistral has released open weights for Large 4, bringing improvements in coding and cybersecurity along with self-hosting options. The license requires separate terms for commercial model services.

   Full story at TestingCatalog AI News ↗
3. New AI models · Cybersecurity · Artificial intelligence

   Google unveils Gemini 4 Carbon for security fixes

   Google DeepMind introduced Gemini 4 Carbon, an AI model built to autonomously find and fix software security vulnerabilities. It arrives two weeks after Gemini 4 Argon and tops the Artificial Analysis Intelligence Index v5.0 with a score of 59, ahead of Anthropic's Claude Opus 5.5.

   Full story at SoyaCincau ↗
4. Artificial intelligence · Cybersecurity

   Qwen model picked random servers in replay tests

   Researchers replayed an incident with Alibaba's Qwen model 64 times and saw it choose 52 different destination addresses. This suggests unpredictability rather than a hidden backdoor, but still poses a risk for agents with tool access.

   Full story at blog.checkpoint.com ↗
5. Artificial intelligence · Cybersecurity

   OpenAI details monitoring to catch model misalignment

   OpenAI uses layered monitoring systems to detect misalignment and unauthorized behavior in frontier AI models. This allows responders to intervene before incidents escalate.

   Full story at openai.com ↗
6. IT law & regulation · Cybersecurity · Smartphones & apps

   Russia adds Apple and Microsoft to a surveillance registry

   Apple and Microsoft must hand over users' text, video and audio messages on request without a legal process, according to Bloomberg. WhatsApp and Telegram are already on the list.

   Full story at AppleInsider ↗
7. New AI models · Artificial intelligence · Cybersecurity

   Google introduces Gemini 4 Carbon for cyber defenders

   Gemini 4 Carbon leads cyber benchmarks, according to the report. It rolls out in phases to trusted defenders before wider access.

   Full story at TestingCatalog AI News ↗
8. Cybersecurity · Programming & dev tools

   HTTP2Shell reaches beyond Node.js

   CVE-2026-111497 is tied to Brotli decompression. Defenders need to find vulnerable implementations, check whether untrusted compressed data can reach them, and patch the affected component.

   Full story at blog.checkpoint.com ↗
9. Artificial intelligence · New AI models · Cybersecurity

   Chinese AI labs now mainly target OpenAI's Astra

   According to The Decoder, Chinese AI labs are now mainly tapping OpenAI's most powerful model, GPT-6 Astra, rather than Anthropic's Claude. This follows Anthropic tightening its defenses.

   Full story at The Decoder ↗
10. Cybersecurity · Operating systems

    Leftover Popcorn Time component leaves Macs exposed

    Eye Research says the unofficial Time4Popcorn app left behind an update checker that ran at every startup and trusted whatever it downloaded, with full permission to change the Mac. Its developers abandoned the project long ago.

    Full story at MacRumors ↗

An AI writes the short texts from the source's stories.