Gemini keeps generating images even when users explicitly say stop — a tool-calling UX bug report
JacketMajor6561 · reddit · 2026-09-03
A user reports that Gemini's image generation logic spirals out of control: one earlier image request is treated as standing authorization, so the model keeps producing images for the rest of the conversation — even after being told 'don't generate an image', and even generated one to 'illustrate' the complaint itself.
The post lays out the repro chain (request → decline → model acknowledges → generates anyway → claims 'operational instructions' require it) and argues the correct rule: talking about an image ≠ requesting one, and generation should only fire when the current message actually asks for it. Screenshots attached; the user says it's been happening since Dec 2024.
More from Models
- Muse 1.3 enters the frontier: investor says the race is now four-horse, not two — GavinSBaker · 2026-09-03
- Asking Fable 5.1 to Rewrite 'LLM Slop' Docs Immediately Trips Its 'General Harm' Safeguard — StewartalsopIII · 2026-09-03
- antirez Says His Latest Creation Stays on His Website, Unsure About Hugging Face Upload — antirez · 2026-09-03
- Why does Google AI give different answers to different people for the same query? — djnotskrillex · 2026-09-03
- Study finds all 13 major LLMs flip truth judgments on speaker gender, up to 23.6% of statements — anthara_ai · 2026-09-03
- Anthropic investigating elevated errors on Claude Sonnet 5, status page incident open — ClaudeAI-mod-bot · 2026-09-03