Что защитит наш код от искусственного интеллекта?

Presentation at the Bank of Russia Conference, October 29, 2025 Artificial intelligence is increasingly writing code, and we are looking for ways to protect repositories from low-quality AI code. Slides: https://speakerdeck.com/yegor256/kto-... My Telegram channel: https://t.me/yegor256news (subscribe!) Blog: https://www.yegor256.com Books: https://www.yegor256.com/books.html GitHub: https://github.com/yegor256 (don't hesitate to follow to stay informed) Twitter:   / yegor256   (follow me!) iTunes: https://podcasts.apple.com/us/podcast... SoundCloud:   / yegor256   Yandex Music podcast by yegor256: https://music.yandex.ru/album/31142286 VK Video: https://vk.com/yegor256news #and #LLM #programming #codetree #tests 00:00 – Introduction: Fear of AI and Programmers' Problems 00:27 – Defects and Attitudes to Errors 02:24 – Researching Errors in Open-Source Projects 03:33 – Security Holes (SQL Injection) 04:47 – Dirty Code: Complexity, Duplication, and Technical Debt 07:10 – How AI Exacerbates Problems: Trust and Errors 09:29 – Metaphor: Junior Posing as Senior 10:04 – Research: AI Code Quality and its Impact on Style 12:33 – Data Leaks Through AI Assistants 14:48 – Management Pressure and a Programmer Survey 17:02 – Five Ways to Protect Yourself from AI 17:42 – Method 1: Small Pull Requests 19:53 – Method 2: Test Coverage Monitoring 23:22 – Method 3: Coding Standards (Coding Manifesto) 26:31 – Method 4: Style Checking 31:14 – Method 5: Architecture Description (ArchUnit, README) 36:45 – Conclusions