LLM Digest

Story

arxiv_cs_ai · Nov 5, 2026 · paper

Source brief

Accepted but Wrong: Silent Failures of LLM Code Repairs after API Migrations

arxiv.orgNov 5, 2026
original source linked

In brief

After a runtime or API migration, an LLM-repaired procedure can run cleanly yet return wrong results, and a release check on typical test inputs ships it. We report a preregistered study of 73 tasks covering 60 documented changes in CPython, Node.js and SQLite. Every candidate is

Read the original at arxiv.org →Open in live feed Read that day’s brief

Accepted but Wrong: Silent Failures of LLM Code Repairs after API Migrations

Earlier in this thread 4 items

Launch HN: Laminar (YC S24) – We trained a model to catch silent agent failures

AI in Production: Panellists Highlight Silent Failures, Harness Engineering, and Agent Governance

Where It Breaks: Diagnosing Perception, Memory, and Decision Failures of a Vision-Language Agent in Pac-Man

Grounded but Not Immune: Confirmation Bias in Agent-as-a-judge