The term comes from operations, where it describes how far an incident spreads, and it has become the standard way to size the risk of a code change before deciding how much scrutiny the change deserves. A typo fix in documentation has a blast radius of zero. A migration that alters a shared table, a change to authentication, or an edit to the pipeline that deploys everything else can take down the business, so those changes earn the most careful review regardless of how small the diff is.
Risk-tiered review policies are built on this idea. Changes with a small blast radius and clean automated checks can be merged by machines, while anything with a large blast radius escalates to a person. Sizing the radius is a judgment about consequences rather than about line counts, which is why it is one of the few review decisions that stays with humans.