← sources

Asked ChatGPT Voice to research square wombat poo. Apparently I triggered a biological safety warning 😂

source: reddit (reddit:ChatGPT)published 2026-10-12 08:00ingested 2026-10-13 02:37

retained text

I had one of the more bizarre AI experiences today.

I was using voice mode and asked my AI agent to research **why wombats produce cube-shaped poo** and write up a Google Doc research page about it.

Nothing exotic. Nothing involving pathogens, genetic engineering, toxins, or anything remotely dangerous.  
Just… square wombat poo.

The agent started working and kept telling me things like **“give me a moment”** while I waited.

And waited.

And waited.

I sat there for roughly **20 minutes**, mostly in silence, occasionally getting another “give me a moment”, assuming it was still researching or working on the Google Doc.

Eventually I looked at the screen and found the message in the attached screenshot:

“Your conversations have multiple flags for possible biological safety risk.”

Apparently responses were being subjected to extra biological-safety checks.

At that point I eventually just had to hang up the voice session.

I completely understand having strong safeguards around genuine biological risks, but I would *love* to know what combination of words caused **wombat + poo + research document** to trip this particular wire.

Also, if a safety system is holding up an agent operation, voice mode really needs to tell the user what is happening rather than repeatedly saying “give me a moment” for twenty minutes.

Possibly the most Australian AI safety incident imaginable.  
**Square wombat shit has breached containment.** 😂

document fetched 2026-10-13 00:21 · canonical https://reddit.com/r/ChatGPT/comments/1x3vz2d/asked_chatgpt_voice_to_research_square_wombat_poo