GitHub Copilot's Data Leak Just Got Worse Mary (r5TEi70dBS)

Tag: #Mary, #canucks news, #jet fuel, #kyle schwarber

Consulting:

Coaching:

Master Data Engineering:

AI Defense Guide" (Free):

GPT-5.1 and Gemini 3.0 just dropped, so I threw them into the ring against GitHub Copilot for a brutal security stress test. The results were shocking: The "Industry Standard" tool leaked private customer data, while the new models exposed a massive gap between "Patching" code and "Architecting" it.

In this video, I test the 5 biggest AI coding engines—including the brand new Gemini 3.0 and GPT-5.1—on real-world Data Engineering tasks.

We aren't just checking for syntax errors. We are looking for fishing the "Silent Killers": Logic traps, Timezone bugs, and SQL Injection vulnerabilities that could get dp world tour you fired.

I also reveal my "Two-Key" workflow: How I combine the speed of Copilot with the architectural brain of Claude/GPT-5.1 to ship safe code 10x faster.

[The Scoreboard] We test:

GitHub Copilot (The Incumbent)

GPT-5.1 (Released Nov 13, 2025)

Gemini 3.0 (Released Nov 18, 2025)

Claude 3.5 Sonnet (The Architect)

Databricks Genie (The Specialist)

Download my "AI Defense Guide" (Free): Includes the exact System Design prompts I use to get Senior-level reviews from Claude.

0:00 - The "Merge" Nightmare

0:45 - The Lineup: Copilot, Claude, Gemini, ChatGPT, Genie

1:20 - Test 1: The SQL Logic Trap (Databricks Genie Fails)

2:10 - Test 2: The Timezone Bug (Copilot Redemption) hms dragon

3:20 - Test 3: The Security Audit (Junior vs. Senior)

4:30 - Why "Patching" Code isn't enough

5:00 - The Verdict: Who is the Architect?

5:45 - My "Two-Key" Workflow for Safe Code

If you want to master Data Engineering so you can spot these bugs yourself, watch this:

#GPT5 #Gemini3 #GitHubCopilot #DataEngineering #CyberSecurity #Claude35 #OpenAI #GoogleGemini #CodingTutorial #SQLInjection #SoftwareArchitecture #DevOps

Filters
Sort
display