# XOR encryption

GCHQ recently posted the following JavaScript code on its Instagram and Twitter accounts:

The code contains two messages. The first is represented as a simple numerical encoding. The second is a secret message that has been encrypted, alongside code for decrypting it. Here are some clues to make sense of how this second message has been encrypted.

The message uses a symmetric-key encryption approach, the XOR cipher, that involves applying the exclusive or (XOR) operator to each letter of the message and the key, recycling the key until all characters have been decoded. The secret message is wrapped up in a Base64 encoding, which is a way of ensuring that all its characters are printable letters and symbols, so it’s possible to include the message within the JavaScript as “gNSkYr+VqyGl1Lhko8fqYq7UpGajiuo67w==”.

Here’s a shorter version of the code in R:

```gchq_message <- "gNSkYr+VqyGl1Lhko8fqYq7UpGajiuo67w==" |>
Β  Β  Β  Β  Β         base64enc::base64decode()
gchq_key <- c(0xc6, 0xb5, 0xca, 0x01) |> as.raw()
xor(gchq_message, gchq_key) |> rawToChar()```

(No spoilers here…)

So, the steps to decrypt are:

1. Translate the Base64 encoded message to raw bytes
2. XOR those raw bytes with the key
3. Translate the bytes to ASCII characters so we can read the message

The nice thing about this form of encryption is that the same algorithm does both encrypting and decrypting. So, if you wanted to reply, “No thanks, I’m good” you just do the same in reverse:

1. Translate your ASCII text message to raw bytes
2. XOR those bytes with the key
3. Translate the result to Base64

In R:

```"No thanks, I'm good" |>
charToRaw() |>
xor(gchq_key) |>
base64enc::base64encode()```

This gives “iNrqda7UpGq1mepI4djqZqnarg==”.

Fun! Also, I have a tattoo that uses the same approach, except I used Braille ASCII instead of Base64 to ensure that all the characters were tattooable π

If you’re watching The Undeclared War, look out for the shout out to Base64 too:

But why is the key c6b5ca01? It’s not obviously the letters G, C, H, Q. In decimal, it looks like an IP address, but there’s nothing obvious at 198.181.202.1, and any four 8 bit numbers look like an IP address if you stare long enough.