Mechanistic Interpretability · EP 2 · Jul 13, 2026

AI Is Lying To Your Face — And We Can Prove It

Name: AI Is Lying To Your Face — And We Can Prove It
Uploaded: 2026-07-13T04:58:52.742Z
Description: The newest research in Mechanistic Interpretability, explained.

This week in Mechanistic Interpretability

Papers covered

Sycophancy makes us worse peopledoi:10.1126/science.aec8352
Ambiguous instructions break LLM labelsarXiv:2607.08961
Quantum circuits don't help diffusion modelsarXiv:2607.09108
AI can't escape its own vocabularyarXiv:2607.09560
When can AI analysis be trusted?arXiv:2607.09128
Personalized AI code review worksarXiv:2607.08990

Your field, every week

Get this for your own field.

Sciport turns the newest papers in your corner of the literature into a short weekly video like this one — auto-curated, delivered to your inbox. First episode’s on us.

Generate your first video →