Learned something interesting today:

Git commit ids are sha-1 hashes of the commit object (parent commits, author, committer, timestamp, file tree etc.)

→ They are 40-character hex string, but for ease of use you see only 7–8 characters

Does this mean that all commit IDs generated are unique?

→ yes, there is a possibility of collision because the output size is fixed and the combination of input is finite.
→ The number of total possible SHA-1 hashes are 2^160. The chances of 2 commits being the same hash is astronomically small, but hey, there is a chance.