# A Dark Day for the Internet: Stolen AI Keys, AI Agents Attacking Banks, a Poisoned AI Tool Today's attacks went after the keys and trust that connect apps to AI. Replace your AI keys, cap what they can spend, and install updates before you trust… **Published:** 2026-10-22 · **Author:** Yash Thakker **Source:** https://explainx.ai/newsletter/dark-day-for-the-internet-stolen-ai-keys-ai-agents-attacking --- > **In short:** Today's attacks went after the keys and trust that connect apps to AI. Replace your AI keys, cap what they can spend, and install updates before you trust any "security fix" link. ## Today's Biggest Attack ### Railway Hack and the OpenRouter Breach Claim: Why Attackers Are Stealing AI API Keys, and What to Do On October 21, Railway, a popular service for hosting apps, confirmed that an attacker reached the system holding its customers' secret settings and decrypted some of them. Railway says the target was keys for AI providers such as OpenAI and Anthropic. OpenCode confirmed a similar breach on October 17, and OpenRouter was named in a breach claim it has not confirmed. A stolen key works like a credit card for AI. **What you can do:** If you use AI API keys anywhere, replace them today, set a spending cap on each one, and give every app its own key so one leak stays small. [Source](https://explainx.ai/blog/railway-openrouter-attacks-ai-api-keys-stolen-what-to-do-2026) ## More Attacks Today ### CyberCartel Used Claude, Grok and Kimi Agents Against 400+ Latin American Financial Targets Security firm Straiker says a group called CyberCartel ran a team of AI agents built on Claude, Grok and Kimi against more than 400 banks and payment companies in Latin America, with little human effort. Anthropic has blocked the group. Which banks were hit and how much was stolen is not public. [Source](https://explainx.ai/blog/cybercartel-ai-agents-claude-grok-kimi-400-latin-american-financial-targets-straiker-what-is-confirmed-2026) ### LiteLLM 1.107.0 on PyPI Is Malicious: What the Credential Stealer Takes and How to Clean Up LiteLLM, a popular Python tool for calling many AI models, shipped a version on PyPI and Docker with a hidden password stealer that runs whenever Python starts. It has been removed, and version 1.106.2 is clean. Anyone who installed 1.107.0 should treat the keys on that machine as stolen. [Source](https://explainx.ai/blog/litellm-1-107-0-pypi-compromise-credential-stealer-wheels-what-to-do-2026) ### Microsoft Ties ConnectWise Attack to China-Linked Storm-0940 and AI Tool ARTEX: What Is Confirmed Microsoft says a China-linked group used ARTEX, an open-source tool that lets an AI model run hacking steps, to break into ConnectWise's cloud backup service and reach customers' Microsoft 365 accounts. About 170 organizations were confirmed hit. How much the AI did on its own is not shown. [Source](https://explainx.ai/blog/microsoft-storm-0940-artex-connectwise-cloud-backup-attack-claimed-vs-verified-what-to-do-2026) ## One Useful Checklist ### eMudhra CA Compromise: The "Update Everything" Security Checklist and What a Viral Thread Missed A viral post told everyone to update everything after today's incidents. The real problem behind it is eMudhra, a company whose certificates help browsers decide that a website is genuine. Browser and system updates fix most of it, but tools like pip, uv and Docker images keep their own copy of that trust list. **What you can do:** Update your browser and operating system today. Never install a "certificate fix" from an email or a pop-up, and never click past a browser security warning. [Source](https://explainx.ai/blog/emudhra-ca-compromise-update-everything-security-checklist-what-you-missed-2026) ## Upcoming Workshops ### AI Builder Workshop: Build and Ship with Claude Code ![AI Builder Workshop: Build and Ship with Claude Code](https://explainx.ai/workshops/covers/square/ai-builder.webp) *Live workshop · Nov 15–23, 2026* Four live sessions for product managers, founders, marketers and developers who want to turn ideas into working software with Claude Code. Setup is guided, and no programming experience is needed. [Source](https://explainx.ai/workshops/ai-builder) ### Claude for Work: Become Your Team’s AI Person ![Claude for Work: Become Your Team’s AI Person](https://explainx.ai/workshops/covers/square/claude-for-work.webp) *Live workshop · Nov 28–29, 2026* A two-session live workshop for professionals who want Claude to work like a real teammate, no coding needed. You build a knowledge base, an agent skill, a prompt library and your own Artifacts, and connect Claude to the apps you already use. [Source](https://explainx.ai/workshops/claude-for-work) ## Reader poll After today's attacks, what will you do first? - Replace my AI API keys - Update my browser and phone - Check the Python packages I use - Nothing yet ## Sources - [Railway Hack and the OpenRouter Breach Claim: Why Attackers Are Stealing AI API Keys, and What to Do](https://explainx.ai/blog/railway-openrouter-attacks-ai-api-keys-stolen-what-to-do-2026) - [CyberCartel Used Claude, Grok and Kimi Agents Against 400+ Latin American Financial Targets: What Is Confirmed](https://explainx.ai/blog/cybercartel-ai-agents-claude-grok-kimi-400-latin-american-financial-targets-straiker-what-is-confirmed-2026) - [LiteLLM 1.107.0 on PyPI Is Malicious: What the Credential Stealer Takes and How to Clean Up](https://explainx.ai/blog/litellm-1-107-0-pypi-compromise-credential-stealer-wheels-what-to-do-2026) - [Microsoft Ties ConnectWise Attack to China-Linked Storm-0940 and AI Tool ARTEX: What Is Confirmed](https://explainx.ai/blog/microsoft-storm-0940-artex-connectwise-cloud-backup-attack-claimed-vs-verified-what-to-do-2026) - [eMudhra CA Compromise: The "Update Everything" Security Checklist and What a Viral Thread Missed](https://explainx.ai/blog/emudhra-ca-compromise-update-everything-security-checklist-what-you-missed-2026) - [AI Builder Workshop: Build and Ship with Claude Code](https://explainx.ai/workshops/ai-builder) - [Claude for Work](https://explainx.ai/workshops/claude-for-work) --- From [explainx.ai](https://explainx.ai/newsletter). Read online, comment and subscribe: https://explainx.ai/newsletter/dark-day-for-the-internet-stolen-ai-keys-ai-agents-attacking