

Inside the DeepSeek AI existential crisis, Chinese 'backdoor' in medical devices
14 snips Jan 31, 2025
Delve into the existential controversies surrounding DeepSeek AI as it competes with established models like OpenAI. Explore the troubling privacy issues and geopolitical implications tied to its Chinese origins. The discussion shifts to alarming backdoor vulnerabilities in medical devices, exposing serious risks in healthcare. Unpack the constant threats from phishing attacks targeting influencers and the VPN industry's innovative solutions against censorship. Lastly, uncover the challenges in cybersecurity regulations and the ethical dilemmas in handling sensitive information.
AI Snips
Chapters
Transcript
Episode notes
DeepSeek's IOC Extraction Prowess
- Costin Raiu found DeepSeek to be superior to ChatGPT for extracting IOCs from blog posts.
- It accurately extracted all IOCs, formatted them nicely, and correctly identified threat actors.
OpenAI's Reasoning Models
- Juan Andres Guerrero-Saade emphasizes OpenAI's reasoning models are superior for complex tasks.
- These models use an agent approach, running multiple queries to formulate comprehensive answers.
Meta's Open-Source Strategy
- Meta's open-sourcing of their model weights challenged OpenAI's first-mover advantage.
- This strategy devalued the cost of training models and leveraged open-source innovation.