Could AI Secretly Inject Backdoors into Your Code? | Eve Cogan & Ulysses Maclaren

Chinese AI models can offer powerful capabilities at a much lower cost, so why are many Western organisations still reluctant to adopt them? Eve Cogan and Uly Maclaren discuss the privacy and security concerns surrounding cloud-hosted Chinese models, including where prompts and company data may be processed. They also explore a more speculative concern: that AI-generated software could introduce vulnerabilities or backdoors into a codebase without developers noticing. As AI-generated pull requests become larger and more complex, human review may become less thorough. The discussion considers whether that creates a genuine software-supply-chain risk—and acknowledges that malicious or vulnerable code could also come from Western models. Will the cost advantage eventually outweigh the trust gap? #ChineseAI #AISecurity #Cybersecurity #CodeSecurity #AIModels #SoftwareDevelopment #AICoding #DataPrivacy #TechNews #SSWTV v2 - Marcus