I beg your pardon if I am actually asking a silly question.
There is nothing in the bitcoin code preventing two different wallets from generating the same address by chance. Given that there is almost no limit to how many addresses you can generate from a seed phrase,* then there are certainly two seed phrases out there which, at different derivation paths, will derive the same address. In fact, there are countless seed phrases which will derive countless numbers of the same address.
However, the reason this will never happen is not down to any in-built code or restriction, but down to simple math. Considering only legacy addresses, there are 2160 possible addresses, which is 1.46*1048. Even if there are a 100 billion people, each with a billion wallets, and each wallet with a billion addresses, then we would still only have used up 0.000000000000000007% of all possible legacy addresses.
*There is actually a limit, but it is enormous. A derivation path can have at most 256 levels (one of which will always be the master), and each of the other 255 levels can have 232 - 1 values, if you include hardened paths. This means the total number of key pairs which can be derived from a single seed phrase is (232 - 1)255, which works out to somewhere around 2.5*102456. This number is obviously far in excess of the number of valid private keys, meaning that any seed phrase taken to its limit is highly likely to produce every valid private key. The reason this doesn't happen is because of the math I have explained above. The sun would be dead before you got even a tiny fraction of the way through the process.