🤖 AI in SOC - Accelerated Defense Response
This post discusses the application of AI in defense cybersecurity operations. It highlights how AI can create synergies within a Security Operations Center (SOC) to improve response times.
Key Points:
• AI applications in defense cybersecurity are being developed.
• The goal is to enhance SOC capabilities.
• AI helps achieve faster incident response.
💡 Malware Analysis - New Book Release
This article announces the availability of "The (Un)Natural History of Malware." The book is published by No Starch Press.
Key Points:
• A new book on malware, "The (Un)Natural History of Malware," is now available.
• The book is published by No Starch Press.
🔗 Resources:
• No Starch Press ↗ - Publisher website for malware book.
🤖 AI Coding Agents - Data Privacy and Training
This discusses the current state of data privacy with AI coding agents. It highlights concerns regarding Zero Data Retention (ZDR) and default data training policies for personal accounts.
Key Points:
• Many personal AI accounts lack Zero Data Retention (ZDR).
• "Do not train on my data" is not the default setting for AI agents.
• These data privacy discussions are still ongoing despite AI agent maturity.
🤖 Autonomous AI Pentesting - Data Visibility Concerns
This addresses the core concerns surrounding autonomous AI pentesting. The debate focuses on AI agents accessing internal network environments and sensitive data.
Key Points:
• Concerns exist regarding AI agents viewing internal environments during pentesting.
• Data visibility for autonomous AI is a central debate point.
🤖 LLM Security - SSRF in LMDeploy Image Function
This details a Server-Side Request Forgery (SSRF) vulnerability found in the LMDeploy LLM image function. The vulnerability allows an attacker to make requests from the server's perspective using a crafted image URL.
Key Points:
• LMDeploy's LLM image function is susceptible to SSRF.
• A crafted image_url can force the server to fetch internal resources.
• The example targets AWS EC2 metadata service.
🚀 Implementation:
- Construct a JSON payload with
image_urlpointing to an internal resource. - Send the payload to
TARGET/v1/chat/completionsusingcurl. - Observe if the LLM backend attempts to fetch the specified internal URL.
🤖 LLM Security - Additional SSRF Information
This provides an additional resource related to Server-Side Request Forgery (SSRF) vulnerabilities in Large Language Models. It suggests further reading on the topic.
Key Points:
• Further information on LLM SSRF vulnerabilities is available.
🔗 Resources:
• X Post ↗ - Related discussion or details on LLM SSRF.
🤖 LLM Data Training - Open Source vs. Proprietary Practices
This post raises a question regarding data training practices in Large Language Models. It suggests a potential distinction in how open-source models handle user data compared to proprietary ones.
Key Points:
• Differences may exist in data training policies between open-source and proprietary LLMs.
• The discussion concerns how user data is utilized for model training.
🔗 Resources:
Image
💡 Cybersecurity Sanctions - EU Targets Russian GRU Hackers
This reports on sanctions imposed by the European Union against Russian GRU military hackers. These sanctions are a response to their involvement in various cyberattacks.
Key Points:
• The EU has sanctioned Russian GRU military hackers.
• Sanctions are a direct consequence of their cyberattack activities.
🔗 Resources:
• BleepingComputer Article ↗ - News on EU sanctions against Russian GRU hackers.
💡 Blockchain Security - 2026 Mid-year AML Report
This references the "2026 Mid-year Blockchain Security and AML Report." The report is available on GitHub and covers security and anti-money laundering aspects of blockchain technology.
Key Points:
• A "2026 Mid-year Blockchain Security and AML Report" is available.
• The report covers security and anti-money laundering in blockchain.
🔗 Resources:
• GitHub Repository ↗ - Repository containing the Blockchain Security and AML Report.
Image
⭐️ Support
If you liked reading this report, please star ⭐️ this repository and follow me on Github ↗, 𝕏 (previously known as Twitter) ↗ to help others discover these resources and regular updates.