ChatGPT: My Cybersecurity Guru for Google Verification

I’ve been using ChatGPT for everything from work tasks to brainstorming names for my son, but this particular use case truly impressed me. Imagine having a patient cybersecurity expert right by your side.

For several months, I’ve been building a web app to summarize meetings and generate transcripts with voice recognition using LMMs. It’s still in the testing phase — feel free to try it — but today, I want to focus on a different aspect of the development journey. My app’s cornerstone is its integration with big-name video conferencing platforms like Zoom, Teams, and Google Meet. While Google Meet integration is complete, Google’s verification stamp is still pending, which might scare users.

To remove this unverified tag, Google mandates a verification process they claim lasts six weeks. Part of this is a code analysis to find any security vulnerabilities. To a self-taught developer like me, this was scary. After all, I’m someone who built this project to improve his coding skills — not as a security expert. If the app takes off, I’ll recruit a security maestro.

Navigating the Scanning Maze

Running my project through a code analyzer for the first time was stressful. I lacked an experienced dev’s eye for spotting security flaws,​​ and I wondered about the number of potential issues in my code. The answer? A whopping 37 errors. 🙃

At first, the scan report was quite demotivating. I stepped away to clear my head. After some downtime, I returned, determined to tackle the errors. The report detailed the issues and suggested solutions, but they weren’t exactly crystal clear. That’s where ChatGPT came in.

A bit of context — I’ve been chatting with ChatGPT about my project for months. It’s got the context of what I’m building including the code base.

Prompting the error information from the scan report together with the corresponding code snippets to ChatGPT was a game-changer. It simplified complex cybersecurity concepts and offered tailored solutions to my code.

Take CSRF as an example — what the heck is CSRF attacks? But ChatGPT broke it down for me, providing both insight and code fixes.

This was a breakthrough. I found myself learning and fixing issues, and with this momentum, I navigated through all 37 errors in one afternoon.

Not Quite Done Yet

The code passed the scan, leaving me with a great sense of accomplishment. Yet, one more challenge awaited — the final survey. I thought, “Surely, a survey can’t be that extensive, right?”

This “survey” was more like a marathon — with 12 sections, each presenting security questions that required detailed answers.

Initially scared, I still had the hype of my coding marathon and ChatGPT’s contextual expertise. So, I buckled down and tackled question one.

Once again, ChatGPT was my hero, explaining questions I didn’t grasp and providing informed responses. For instance, one question on encrypted data authentication had me puzzled — HMAC, cipher modes…what? But ChatGPT reminded me of JWT’s signature feature that I was using and wrote the perfect justification.

Reflections on the Journey

Certainly, ChatGPT can sometimes miss the point, but in specific scenarios, it significantly boosts productivity. Tasks that could have taken days or weeks of cybersecurity research were completed in just a few hours.

Launching the product and see actual user engagement is my main focus, not getting bothered with advanced security features for a product that might not even have users. I’ve even intentionally designed my project to handle minimal sensitive data and auto-delete recordings post-summarization to avoid security complications.

With ChatGPT, I’ve navigated the cybersecurity labyrinth and come out the other side, ready to launch. Let’s see where this goes!