Multisig: no single point of failure
A single-key wallet has exactly one point of failure: compromise that one key (via a phishing signature, a leaked seed, a hacked device) and the attacker has full, immediate, irreversible control. A multisig wallet requires M-of-N independent signatures to approve a transaction, for example 3-of-5 distinct keys held by different people or devices, no single one of which can move funds alone. Compromising one signer accomplishes nothing without also compromising enough of the others, ideally kept on different devices, in different locations, or held by different people entirely, so a single phishing campaign or device compromise can't reach a majority at once.
This is precisely why treasuries, DAOs, and serious protocols run multisig for anything beyond trivial amounts: it converts a single catastrophic failure into a coordination problem an attacker has to solve across multiple independent, differently-secured targets simultaneously. The tradeoff is operational friction: every transaction needs multiple people to review and sign, which is slower but is the entire point, since it also builds in a human review step that a lone signer's wallet doesn't have.