Trust but Verify: Coding Securely in the Age of AI | My Tech Talk at скачать в хорошем качестве

Trust but Verify: Coding Securely in the Age of AI | My Tech Talk at

Is your AI-generated code functionally correct but secretly vulnerable? In this talk from the DevCongress January 2026 Meetup, I explore "The Vibe Coding Trap" and why 45% of AI-generated code contains vulnerabilities. We dive deep into the golden rule of AI development: Treat AI code like input from an untrusted user. 🔍 What we cover in this session: The Trap: Understanding why functional correctness does not equal security. Real-World Scenarios: 🐍 Python: The "Gullible Endpoint" and command injection risks. 📜 JavaScript: Handling "Hallucinated Dependencies" in your imports. #️⃣ C#: Spotting "Hardcoded Secrets" in connection strings. The Checklist: A 5-step "Trust but Verify" guide including Input Validation, Secret Hygiene, and Automated Security tools. Remember: "AI writes the draft. Humans edit the final copy." 👨🏾‍💻 About the Speaker: Isaac Amankwaah Anane is a Software Engineer passionate about mentorship, software architecture, and career coaching. Portfolio: https://isaacanane.com/ LinkedIn:   / isaac-amankwaah-anane   YouTube: @IAmIsaacAnane 📅 Event: DevCongress Meetup (January 2026) - Hosted at MEST Africa. TABLE OF CONTENT 00:00 - Introduction & Speaker Bio 00:45 - The Vibe Coding Trap 01:30 - The Golden Rule: Trust AI Like Untrusted Input 02:15 - Stat: 45% of AI Code is Vulnerable 03:00 - Scenario 1: The Gullible Endpoint (Python) 04:30 - Scenario 2: Hallucinated Dependencies (JavaScript) 06:00 - Scenario 3: Hardcoded Secrets (C#) 07:30 - The Trust but Verify Checklist 09:15 - Automated Security & SAST Tools 10:30 - Conclusion: Humans as Final Editors 11:15 - Q&A Session #AI #CyberSecurity #SoftwareEngineering #DevCongress #Coding #Python #CSharp #JavaScript #Copilot #SecureCoding