What can possibly go wrong? Cybersecurity awareness month: 👇

AI Cyber Podcast is back! My guest this week, used to submit 10 to 20 bugs in a good month. Her AI research system found 195 in 3 weeks, more than her entire manual bug bounty career. Most of them were in software with no bounty program, which means nobody paid her for a single one. Funny thing is, @Thezinnekalu was enrolled in medical school but training to be a medical wasn’t going well for her. So she quit! However, she loved her time at a four-week Microsoft program on Linux and also enrolled and graduated from the CyberGirls Fellowship hoping to kickstart her career in #cybersecurity. However, she hit another snag when she couldn’t land a job. So Ezinne Kalu started finding vulnerabilities in other people's software to prove she was a great talent. 4 years later she now runs what she calls a one-lady security lab and holds assigned CVEs. In this episode of AI Cyber podcast, Ezinne Kalu walks us through her vulnerability research system; Claude Code as the harness, a second brain built from Andrej Karpathy's pattern, everything stored in GitHub and managed in Obsidian. She discussed her first assigned CVE; CVE-2026-79653 which is a path traversal in Eclipse SW360, software running inside more than 30,000 organisations including Siemens. It allowed arbitrary file write with system privileges, which means remote code execution. It had been sitting in the code for at least six released versions. Ezinne Kalu found it, reported it, and it is now patched. She also shares how she runs her AI research system cheaply with six practical tips for anyone building this themselves. She said her system now produces more findings than she can personally disclose….isn’t that mind blowing?! So what happens when the constraint on security research moves from finding bugs to reporting them, and what does that means for every maintainer about to receive a flood of reports that are not AI slop? Don’t miss this week’s episode of AI Cyber Podcast for any reason! Subscribe to our channel on YouTube and turn on notification. youtube.com/@aicybermagazine…

Nigeria has become cloud-first or cloud-by-default! This policy mandates every Ministry Department and Agency (MDAs) designing digital systems, to do cloud deployment. Infact, for MDAs to opt now of cloud deployments, they'll need a published, time-bound exemption, assessed by NITDA against published criteria and recorded in a public register.

Nigeria has recently unveiled a National Digital Cloud Policy with $750Million expected to be invested over the next 24 months. We should all be paying attention! I have read it in detail and will be sharing in a 3-5 part series, what the policy actually does, what can go wrong from a cybersecurity stance, the opportunities for Nigerians and the talent Nigeria needs urgently. This is part 1 - what the policy actually does.

Remember my post about the new program? It's time... We ( @cybersafehq) are collaborating with @SANSInstitute for this one; and if you meet the eligibility criteria, you don't want to sleep on it! Catch up with the details in the video

AI isn’t inventing new vulnerabilities, it’s multiplying the old ones. In this episode of AI Cyber Podcast, Roeland Delrue explains that LLM agents are trained on human mistakes, so they replicate them faster, producing larger, harder-to-review pull requests that amplify supply chain risks. Watch the full chat on @aicybermagazine👇 youtu.be/N-b-050MQr4