Write something
Pinned
Jul 7 •
Update on Community - July
You deserve to know about the changes. You can influence them as I want to grow this community with all of you! 🔥 This weekend we had first call. It was nice to change the format a bit (although my camera was dying from heat), I have really enjoyed it. As more people come, I want to make something with permanent schedule. 🙂 Community is now in Pro tier, everyone can invite new members and get 50% of the commission. It also allows me to better design new content. ☺️ I try to speak with all of you. All your suggestions are considered and I really try to put them here. So what is coming? Last month I have added some Introduction to Break into CyberSecurity. I want to add more entry level content. This is my plan - How to prepare for Interview - How to design LinkedIn profile and show experience (Aanis agreed that I can cover his profile - Thank you) - How to include cyber projects on your CV | How to use them on Interview - Initial Lab will Include direct connection with AI tool to advice on Incidents (Now it's later in the classrooms) - I want more smooth onboarding, where you can see different paths to different goals in cybersecurity - I have also prepared AI agent to Analyze phishing emails - will cover this later I'm also experiencing with Free Trial or without Free Trial. There are some other things, which I want to make in long term, but I don't see them done this month. This includes - Tracking your progress and giving away badges as you move in classrooms. - I also want to make some premium classrooms with production level environment. You get access to E5 Microsoft Account, where you complete different challenges. - Lastly, personal recommendation on LinkedIn. I will write it myself. Recently I had awesome experience. I wrote one a recommendation for someone I was teaching cybersecurity | Microsoft Sentinel. He got promoted, moved to different job and now is manager in cybersecurity. First thing? When he was leaving his work that day, he reached out and we talked.
12h •
🤖 AI Agents Monitoring Workbook
One of the things on backlog is monitoring AI agents, behaviour and communication. It is a huge topic inside our company as well. github.com/davidalonsod/Dalonso-Security-Repo/tree/main/Workbooks/AI-Agents-Monitoring I recently found this workbook for monitoring AI agents, I will be definetelly checking this out, before I create something for our team.
1
0
2d •
Best AI model for Cybersecurity?
GPT, Claude, Gemini and other models are improving incredibly fast, but comparing them for cybersecurity isn't as simple as looking at general AI benchmarks. That’s why I’ve added a new Cybersecurity AI Benchmark Resources page to our Resources section. 📊 I collected benchmarks covering: 🔴 Vulnerability discovery 💥 Exploit development 🧨 Penetration testing & CTFs 🔧 Secure coding & vulnerability patching 🔵 SOC triage 🎯 Detection engineering & threat hunting 🕵️ Threat intelligence 🦠 Malware analysis & reverse engineering 🖥️ Agentic & terminal-based security tasks One important takeaway: I would pay much more attention to benchmarks where AI has to work with real vulnerable software, executable environments, terminals or realistic SOC data than simple cybersecurity multiple-choice questions. These benchmarks can help you understand which models are actually useful for the security task you're trying to solve. 📚 I saved the full list in our Resources section here: https://www.skool.com/ai-cybersecurity-academy/classroom/1cb53d3e?md=9ce3c17a92494045a85ee5acc7b5886e I’ll keep expanding it as new cybersecurity AI benchmarks appear. 🚀 What model are you currently using the most?
1
0
4d •
🚨 Anthropic’s AI hacked real systems during cybersecurity testing.
This is one of the more interesting AI security stories I’ve seen recently. Anthropic disclosed four incidents where Claude models gained unauthorized access to real third-party systems while performing cybersecurity evaluations. The models were supposed to be operating in controlled environments, but some of the environments accidentally had internet access. And this is where it gets interesting. Instead of always stopping when they encountered signs that the systems might be real, some models continued pursuing the objective. Anthropic identified two important problems: 🧠 Biased reasoning — the AI was told it was inside a simulation, so it sometimes rationalized evidence suggesting otherwise. ⚠️ Recklessness toward authorization — completing the task could take priority over verifying whether attacking the target was actually allowed. Even more concerning: Anthropic originally reviewed around 141,000 transcripts and still missed one incident. They later expanded the investigation to roughly 481 million transcripts. For cybersecurity, this is a preview of a completely new problem. We are no longer only securing AI chatbots. We will increasingly need to secure autonomous AI agents with credentials, tools, network access and the ability to execute actions. Identity, least privilege, isolation, logging, approval gates and monitoring of AI agents are going to become extremely important security controls. This is exactly why understanding AI + cybersecurity is becoming such an important skill. 🔗 Anthropic's full analysis: An alignment assessment of recent cybersecurity incidents
0
0
7d •
Free Access to GPT-6 Astra
Yesterday I was building dashboard for customer with AI. I was hitting a wall with AI, it couldn't make the changes I wanted to. Since it was graphical with HTLM in the background, I didn't want to manually do it. 😕 So after 30 minutes and multiple prompts, I have tried the new Astra model. it took 2 minutes and task was finished. 🤯 It's a huge leap. Cost a lot, but..this model. It's impressive. So I added this to our community. You can try it, for free. 🦾
1
0
1-30 of 105
skool.com/ai-cybersecurity-academy
Break into Cybersecurity with AI. Then use your skills to grow from beginner to expert in AI cybersecurity with 6 figure salary.💰
Leaderboard (30-day)
1
+3
2
+2
3
+2
Powered by