Sponsored
$5/week ↗
My Dot booked an appointment, built my birthday shopping list, and asked to pass CAPTCHAs for me, image 1 of 1
Projects

My Dot booked an appointment, built my birthday shopping list, and asked to pass CAPTCHAs for me

added

I started testing @openai ChatGPT new Dot feature with some VERY ordinary tasks...and then it pretended to be human. Book an appointment. Build me a birthday shopping list. Help me sort a few things while I was driving to Sydney. Within about 20 minutes, it asked me something I genuinely wasn't expecting: Could it complete an 'I am human' check for me? Then it asked whether I wanted to let it handle future CAPTCHAs too. The shopping list came back. My booking was accepted. In another task, Dot told me it could use a card already saved in the browser environment to pay a deposit if I approved it.

First-hand test by @ai_herway (Dr Nici Sweaney, AI keynote speaker): she gave her ChatGPT Dot ordinary tasks and got concrete results — an appointment booked (booking accepted), a birthday shopping list delivered, and errands sorted while she was driving. Along the way the Dot asked to complete an 'I am human' check and whether it could handle future CAPTCHAs — she approved to see what happened and it saved a persistent permission rule. Screenshots in the carousel show the permission request, the saved rule, and downstream results. Note her own caveat: the screenshots don't prove exactly how the human check was handled — she flags the trust gap between cute interfaces and real delegated capability.

See all 285 →