Proxy

When you visit a website, your device normally sends a request directly to the computer where that site is hosted. A proxy inserts itself into this connection. Your request first goes to the proxy, and it then fetches the page for you. The English word literally means stand-in or representative, and that is exactly its job. The operator of the site sees the proxy’s request, not your own. You can think of it like a messenger who picks up a package and brings it to you.

The stand-in as bouncer and cloak of invisibility

The most obvious benefit is anonymity. Every device on the internet has an address, the so-called IP address. It roughly reveals which country and which provider someone is coming from. Anyone who goes through a proxy leaves the proxy’s address at the destination site instead of their own. That’s why, for instance, journalists in countries with censorship use such detours.

Companies and schools deploy proxies for the opposite reason: as a checkpoint. When all traffic runs through one device, rules can be enforced there. Certain sites get blocked, others get logged. A school network where some websites won’t load is almost always working with such a filter.

For AI companies, the point is different. Anyone collecting huge amounts of text from the web to train a language model makes millions of requests. Many websites automatically block access when too many requests come from the same address. Spread across thousands of proxies, the traffic looks like ordinary visitors. This exact issue has been the subject of legal disputes for years.

The path of a request via the detour

Technically, a proxy is a program on a server that receives requests and forwards them. Your browser is set up so that it no longer goes out directly but sends everything to this one address. The proxy then opens its own connection to the destination and sends the response back to you. Nothing changes on your screen, only the path is longer.

Many proxies cache frequently requested content. This is called caching. If a hundred people on the same network request the same image file, the proxy fetches it only once and then delivers it from its own storage afterward. This noticeably saves time and data volume.

It’s important to distinguish between two types. A forward proxy sits on the users' side and protects or controls them. A reverse proxy sits in front of a web server and, conversely, protects it. It distributes incoming visitors across multiple machines so that none gets overloaded. A common misconception, by the way, is that a proxy automatically encrypts everything. It doesn’t – it actually knows your traffic, so you have to trust it.

Proxies in school networks, streaming, and the cloud

In everyday life, you encounter proxies mostly without noticing. Almost every larger website today runs behind a reverse proxy, often from providers like Cloudflare. These services also intercept attacks in which someone tries to take down a site with a flood of requests.

The technology becomes visible with streaming. Some shows are only available in certain countries because the rights were sold that way. Anyone going through a foreign proxy sees a different selection. Providers now recognize and block such addresses fairly reliably.

In tech news, the term also comes up in connection with AI services. Companies frequently place their own proxy in front of language models like ChatGPT. There, costs get tracked, sensitive data gets filtered out, and requests get logged. Related, but not the same, is a VPN: it reroutes all of a device’s traffic in encrypted form, whereas a proxy usually only affects individual programs.