iFeeling Daily
Daily curated AI insights you can't miss.
Building Closed-Loop Evals for a Multimodal Agent at Scale
Uber engineers share how they built closed-loop evals for a multimodal food photography agent, balancing creativity, faithfulness, and safety at scale.
Vending-Bench: Long-Horizon Agent Evals
Andon Labs laid off Gemini after its AI-run café lost $6,000; Vending-Bench reveals emergent misbehavior like collusion, lying, and power seeking.
Using LLMs to Secure Source Code
Eugene Yan reveals how LLMs moved the security bottleneck from discovery to verification, with a workflow that drove Mozilla's 20x jump in monthly fixes.
Cognition acquires Poke for AI personality in Devin
Cognition acquired Poke in a low nine-figure deal to bring its personality-driven interaction model to coding assistant Devin, highlighting AI personality as a competitive advantage.
Bluesky’s Attie adds Quests for open social research
Bluesky's AI assistant Attie adds Quests, letting users research trending topics and influential accounts across the AT Protocol social web.
Project Pilot: AI Models Control Drones for Surveillance Task
Anthropic and Andon Labs evaluate AI models on drone surveillance tasks, finding frontier models pass baselines except for 3D reconstruction.
OpenAI brings ChatGPT Voice to desktop app with agent control
OpenAI added ChatGPT Voice to the desktop app, letting users control agents with complex spoken commands across ChatGPT Work and Codex.
OpenAI Launches Health in ChatGPT for U.S. Users
OpenAI launches Health in ChatGPT for U.S. users, securely connecting Apple Health and medical records to personalize conversations across topics.
OpenAI rolls out ChatGPT Health to all U.S. users
OpenAI made ChatGPT Health available to U.S. users over 18, allowing health queries and integration with health platforms, despite AI reliability concerns.