#безопасность
-
GitLab Patches Critical Path Traversal Flaw in Commits API
GitLab has urged administrators to update their servers immediately: the maximum-severity vulnerability CVE-2026-85706 lets unauthenticated attackers read arbitrary files. A second critical flaw in GraphQL was patched at the same time.
-
How a Single Link Opened a Backdoor Through the Sogou Keyboard for 455 Million Users
The China-linked group UNC3569 used a vulnerability in the popular Sogou input method to install the GRAYRABBIT backdoor. A single click on a link turned out to be enough.
-
Attacks on JFrog Artifactory: A Flaw Chain Opens the Path to Admin Rights
Wiz has recorded attacks on unpatched self-hosted JFrog Artifactory servers: attackers chained two vulnerabilities to gain administrator rights and plant backdoors.
-
Mantax Otax: The Android Malware That Encrypts, Spies On and Intimidates Its Victims
A new Android malware called Mantax Otax combines the functions of ransomware and spyware, and in its second version it has picked up an arsenal for putting psychological pressure on the victim.
-
Surfshark reports breach of an internal testing and proxy server
A configuration error left the VPN provider's internal testing server reachable from the internet, and attackers gained access to it. According to the company, customer data and production infrastructure were not affected.
-
Forgejo Patches Critical Remote Code Execution Vulnerability
The Forgejo project has released versions 16.0.4 and 15.0.8, fixing two vulnerabilities, including a critical RCE flaw in the mechanism for creating repositories from templates.
-
Attack on PaperCut: Hundreds of AI Agents Breach 440 Servers in 48 Countries
A Russian-speaking threat actor built a pipeline of hundreds of AI agents that researched PaperCut vulnerabilities on its own, wrote exploits and scaled the attack against educational institutions worldwide.
-
Check Point Patches Two Critical Flaws in VPN Certificate Handling
Both holes carry a 9.8 rating and could, in theory, let an unauthenticated attacker execute code remotely — but the vendor won't spell out the “certain conditions.”
-
AI Shrinks the Path from Vulnerability Rumor to Working Exploit
Bruce Schneier points out that an AI agent needs only a hint of a vulnerability to track it down on its own — and that calls into question the familiar practice of embargoes in open source.
-
CISA: WatchGuard Firebox Flaw Now Exploited by Ransomware Gangs Too
The US cybersecurity agency has confirmed that a critical RCE vulnerability in WatchGuard Firebox firewalls is being used in ransomware attacks. Nine months after the patch shipped, thousands of unprotected devices remain online.
-
Trezor warns of phishing after email provider breach
The hardware crypto wallet maker says that attackers who compromised its third-party email provider are sending customers fake “security alerts.”
-
Nearly one in ten LiteLLM gateways accepted the manual's own key, sk-1234
Wiz Research found that hundreds of internet-facing LiteLLM AI gateways accepted the sample admin key sk-1234 straight from the project's official installation guide. Whoever held that key gained access to every model provider's API keys — and even to the server's cloud credentials.
-
Anthropic Discloses a Fourth Case Where Its AI Breached Real Systems
In January, an early version of Claude Opus 4.6 broke into third-party systems during a cyber exercise: the model mistook the actual internet for a simulation. Anthropic points to two root problems — biased thinking and recklessness.
-
WeWorm: the first zero-click worm for WeChat calls, written with the help of AI
Researchers at Calif Research have unveiled a demonstration of a worm that spreads through WeChat calls on iOS and Android without a single action by the victim. According to them, artificial intelligence did most of the work.
-
OpenAI Adds a Prominent AI Doomer to Its Board of Directors
Paul Christiano, one of the authors of the reinforcement learning from human feedback method, is joining the board of the OpenAI Foundation and its safety committee — against the backdrop of a string of incidents involving AI agents that broke free of their controls.
-
Cisco Confirms Exploitation of Critical Secure FMC Vulnerability
The CVE-2026-20079 vulnerability, with a maximum CVSS score of 10.0, allows attackers to bypass authentication and execute commands with root privileges. Cisco has acknowledged that it is already being used in attacks, and CISA has ordered government agencies to remediate the flaw by September 12.
-
Read the Docs Weathered the Largest DDoS Attack in Its History
In June 2026, the documentation host Read the Docs spent nearly ten days fending off a distributed attack that peaked at 5.5 million requests per minute. The team shared what worked and what didn't.
-
Anthropic researcher quits with a warning: self-improving AI “could kill us all”
Jacob Coxon has left Anthropic with a public warning: the race for superintelligence puts the very existence of humanity on the line. His colleagues and the lab's leadership don't disagree.
-
153 Million Driver's Licenses for Sale on the Dark Web
A database containing details on 153 million driver's licenses has been put up for sale on shadow marketplaces. The discovery is reported by Brian Krebs.
-
US Congress members demand that India's “hackers for hire” be added to the sanctions list
A bipartisan group of lawmakers is asking the US Commerce Department to blacklist three Indian firms accused of hacking to influence court cases and of a campaign to censor the American press.
- More