AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
22min chapter
[Crosspost] Adam Gleave on Vulnerabilities in GPT-4 APIs (+ extra Nathan Labenz interview)
The Inside View
Vulnerabilities in GPT-4 APIs and Malicious Code Generation
The chapter explores the specific vulnerabilities in GPT-4's APIs, discussing the implications of fine-tuning the model and the potential for automating harmful attacks. It delves into how even a small dataset can be used to manipulate models for biased responses, highlighting the challenges in controlling these behaviors. The conversation also touches on the risks posed by malicious code generation, emphasizing the competitive disadvantage and difficulty in detecting backdoors in model-generated code.
00:00
Transcript
Episode notes
Get the Snipd
podcast app
Unlock the knowledge in podcasts with the podcast player of the future.
AI-powered
Discover
highlights
Listen to the best highlights from the podcasts you love and dive into the full episode
Save any
moment
Hear something you like? Tap your headphones to save it with AI-generated key takeaways
Share
& Export
Send highlights to Twitter, WhatsApp or export them to Notion, Readwise & more
AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Discover
highlights
Listen to the best highlights from the podcasts you love and dive into the full episode