Catch up, a week or a month at a time.
Every Sunday we publish a weekly review of the episodes that dropped that week, and at the start of each month a deeper monthly review across the whole month. Each review is a permanent, dated snapshot. Browse the latest below, dig into a month, or follow a topic across time.
Latest weekly reviews
AI Papers Week in Review: July 27–August 2, 2026
Inside the Model: Sycophancy, Emotion, and Bias · What Our Measurements Miss: Hype, Culture, and Concepts
AI Papers Week in Review: July 7–13, 2026
Inside the Model: Sycophancy, Emotion, and Bias · Can We Still Watch the Model Think? Oversight and Monitoring · Gaming the Reward: Specification Hacking and Emergent Misalignment · +3 more
AI Papers Week in Review: June 29–July 5, 2026
Teaching Agents to Predict the World · What Agents Remember: Memory, Skills, and Learning on the Job · AI for Scientific Discovery · +9 more
AI Papers Week in Review: June 22–28, 2026
Training and Reinforcement Learning for LLM Agents · Coding and Software-Engineering Agents · When Agents Cause Harm With No Attacker in the Loop · +6 more
AI Papers Week in Review: June 15–21, 2026
Building Deletion Into the Model · Agents on the Attack: Offensive Capability and Adversarial Robustness · When Agents Cause Harm With No Attacker in the Loop · +6 more
AI Papers Week in Review: June 8–14, 2026
Coding and Software-Engineering Agents · Training and Reinforcement Learning for LLM Agents · Evaluating, Serving, and Deploying Agents at Scale · +6 more
AI Papers Week in Review: June 1–7, 2026
Agents on the Attack: Offensive Capability and Adversarial Robustness · When Agents Cause Harm With No Attacker in the Loop · Many Models, One System: Collective Dynamics of Multi-Agent LLMs · +2 more
AI Papers Week in Review: May 25–31, 2026
Training and Reinforcement Learning for LLM Agents · Rethinking Attention, Memory, and Latent Compute · AI for Scientific Discovery · +4 more
Monthly reviews
2026
AI Papers Month in Review: July 2026
What Reasoning Actually Buys: Selection, Recall, and Test-Time Compute · What Agents Remember: Memory, Skills, and Learning on the Job · Many Models, One System: Collective Dynamics of Multi-Agent LLMs · +12 more
AI Papers Month in Review: June 2026
Gaming the Reward: Specification Hacking and Emergent Misalignment · It's the Harness, Not the Model: Scaffolding as the Real Lever · What Agents Remember: Memory, Skills, and Learning on the Job · +15 more
AI Papers Month in Review: May 2026
Inside the Model: Sycophancy, Emotion, and Bias · Training and Reinforcement Learning for LLM Agents · Coding and Software-Engineering Agents · +8 more