Have you ever had to identify images of traffic lights or busses when logging in to a website or to order a product? This is Google’s ReCaptcha trying to determine whether you are human. Ironic, huh? According to Google, ReCaptcha allows them to fight bots on the internet. ReCaptcha often asks you to identify things like traffic lights, cars, and other objects, before giving you access to the website you were trying to enter.

This is essentially a gamified way of training Google's artificial intelligence vision applications by outsourcing the labeling of objects to internet users. Every time you tell the system you recognize the traffic lights, you are essentially feeding it information on which it can train its image labeling models. These AI vision models have the potential to be used in lucrative applications like object recognition or self-driving cars, hence the images of traffic lights and cars.

Next to that though, there is a bigger risk associated with ReCaptcha, which has recently even caused Cloudflare to stop using the service [1].

To determine whether you are a risky visitor, likely to be a bot, ReCaptcha checks various variables, which include whether you are logged into Google and whether you have a Google cookie installed on your device. If any of these are true, you are identified as non-threat. Now Google is demanding that website administrators should embed ReCaptcha v3 on all pages of their site because that would allow the system to learn how website visitors behave over time. Officially the goal is to generate more accurate risk assessments, but this also opens up the possibility of Google tracking users across all pages with ReCaptcha running in the background [2].