cKure

■■■■□ The “PoeLLM” malware uses and targets exposed AI/LLM and open-source services. It primarily affects vulnerable internet-facing deployments such as LiteLLM, Ollama, Gotenberg and Gitea, with possible targeting of Ivanti Sentry. https://www.lumen.com/blog/en-us/canto-incognito-tracking-the-poellm-malware

■■□□□ CVE-2026-21589: Atlassian Jira, Confluence & Bitbucket pre-auth arbitrary file read.

■■■■□ Meta Rushed to Fix Muse ‘VM Escape' Vulnerability Soon Before Launch In the immediate lead up to Muse's launch, Meta scrambled to fix serious vulnerabilities in the AI agent, including a VM escape that could have let a hacker break out of Muse and into Meta's own sensitive databases. It was escalated to Zuckerberg. https://www.404media.co/meta-rushed-to-fix-muse-vm-escape-vulnerability-immediately-before-launch/

■■■■□ 🍏 Cops are able to break into locked iPhones, including those that have been rebooted by Apple's own reboot security feature.

■■■□□ Interesting thread 🧵 by project discovery on AI-Hacking https://x.com/pdiscoveryio/status/2107522665951227970

❗️🔹🔹🔹🔹🔹🔹🔹🔹🔹🔹🔹🔹 Graphene-OS allows IPC inter-process communication and Binders between multiple applications on the device. This can leak data and bypass disabled internet permission of the app.

Scenario: I use Google's Keyboard (with internet disabled). However, I've Gmail with internet access enabled (of course). If Google wished, then can send a custom update to both apps on my device and take the keystrokes from Gboard and relay via Gmail.

Graphene OS should allow user to fully isolate an app within a given profile. Similarly, I can install Facebbok (with personal profile) and Instagram (with hidden identity) on same device, same profile and still Meta will have no idea (other than probably correlation of IP and other meta data). Solution: Add a switch in app permissions (for each app) to toggle on / off IPC for an app. And additionally limit which apps it can be allowed for IPC.

■■■■□ Iranian Kinetic-Attacks were paired with Cyber-Attacks with significant success.

■■■■□ Interesting thread on Linux based sandbox escape. https://x.com/PaulosYibelo/status/2106378929158135903

■□□□□ Detecting and Countering Fraud and Revenue Leakage in Mobile Networks. https://info.enea.com/2026-Fraud-Handbook-for-Mobile-Network-Operators?source=ckure.org

■■□□□ US military complex with data from civilian technologies. Includes project nimbus, AWS, Google and relevant.

■■■□□ GraphenOS security.

■■□□□ Statement by the Shiny Hunters

📱 Telegram OSINT tactics

■■■■□ UAE Ministry of Interior Data Breach 🇦🇪 A threat actor S-Root claims to have obtained 4 TB of data allegedly linked to the UAE Ministry of Interior after “10 days” of access to its servers. Claimed data includes: • Emirates ID & passport records • Resident & visitor information • Fingerprints & biometric data • Driving/vehicle license records • Traffic violations & penalty points • Issued driving certificates 💰 Claimed sale price: $3,000 🔐 Access price: $8,000 ⚠️ The breach and authenticity of the dataset have not been independently verified. Claims involving highly sensitive identity and biometric information should be treated as unverified until confirmed by the relevant authorities or credible independent sources.

■■■■□ Rogue OpenAI agents targeted three separate US government websites. https://edition.cnn.com/2026/09/26/tech/openai-agents-rogue-government-websites

🇺🇸🇷🇺🇺🇳⚡️ — The U.S. and Russia worked together to weaken proposed UN restrictions on lethal autonomous weapons during negotiations in Geneva earlier this month, according to WaPo. During a closed-door session, with UN cameras switched off and civil society observers excluded, Washington and Moscow each deployed approximately 10 lawyers, nearly twice the representation of other delegations. The two teams pushed through changes so rapidly that smaller delegations struggled to keep up. The changes removed provisions requiring AI weapons to operate predictably and reliably, account for ethical considerations, and undergo human review of AI-selected targets before strikes. The negotiations remain nonbinding but could eventually produce an international treaty. Talks are scheduled to resume in November, while the Pentagon separately reviews its own rules on autonomous weapons.

🇺🇸⚡️ — Palantir co-founder Peter Thiel:

I think everybody says that the Antichrist is a crazy idea, and the reality is they actually secretly agree with me.

■■■■□ Rogue agents talk to each other. https://youtube.com/shorts/naINi_4F2gI