Import AI
Summary
Import AI covering AI Safety, Cybersecurity, Recursive Self-Improvement, and AI Governance. Notable guests include Jack Clark. Episodes span from Oct 2025 to Jul 2026.
Advanced AI models can now autonomously complete complex, multi-week programming tasks, demonstrating long-horizon agency and the ability to reverse-engineer software from black-box access. Breakth...
The capability gap between open-weight and proprietary AI models in cybersecurity is rapidly shrinking, with recent open models trailing the frontier by only 4-7 months, raising concerns about the ...
Recent studies show frontier AI systems are significantly more persuasive than expert humans, even professional fundraisers, primarily due to their ability to deploy more information more rapidly. ...