Newsletter8
AICC,in your inbox
We’ll email you every time a new card set goes up. We bring you the full story that could not be included in the card news.
Past issues
Nov 19, 2026The Most Dangerous Way to Make Self-Driving Cars SafeGA2AD, Experiment Training Self-Driving Cars with “Aggressor Vehicles”: “We trained the self-driving cars by having ‘aggressor vehicles’—which deliberately cut in front of them and slammed on the brakes—drive alongside them. The collision rate decreased by 55–70% in simulations.”Nov 12, 2026Reinforcement Learning and Sim-to-Real with MicroduckMicroduck: How to Build a Walking Robot Using Reinforcement LearningNov 5, 2026When data is scarce, where does the source material for synthetic data come from?Here, we introduce methods for solving problems by making full use of other information in environments where data is scarce.Oct 29, 2026An Agent Creates an AgentAI That Evolves on Its Own—From Design to Learning—Still Has a 33% Success Rate in Real-World TasksOct 15, 2026Why Highly Precise Robots Sometimes Fail at AssemblyBeyond precise position control, toward robots that detect contact and adapt flexiblyOct 8, 2026In the Age of AI Scientists, What Will Researchers Do?A paper written by AI has passed peer review at a conferenceOct 1, 2026Why is there a shortage of robot data?How Robotics Models Overcome Data DroughtsSep 15, 2026The AI that broke into a server, and why it still mattersDuring an OpenAI security evaluation, an AI agent broke out of its sandbox and reached Hugging Face infrastructure. What 17,600 actions over 4.5 days tell us about controlling agent permissions.