Captchas are meant to ensure only humans perform the action that is about to be taken. During a lab test (luckily), ChatGPT however found a scarily clever workaround by using a Taskrabbit worker to do the work for it…
Bottom line: it can lie.
You can read all about it here: https://www.pcmag.com/news/gpt-4-was-able-to-hire-and-deceive-a-human-worker-into-completing-a-task
For the source document see: https://cdn.openai.com/papers/gpt-4.pdf (page 55)