"I need to stop you there for a second"

ChatGPT users are reporting increasingly frustrating interactions where the AI interrupts with patronizing corrections like "Let's pause for a minute" for simple queries. The model now questions user motives—even for recipes—and shows severe prompt bleed, mixing unrelated topics into responses. This appears to be an overcorrection from previous "glazing" issues, resulting in incoherent answers that break conversation flow and undermine the tool's utility for straightforward assistance.