A sidechain is not a layer-2. It does not post data or proofs to the base chain, so it does not inherit the base chain's security. It stands on its own validator set, which is often far smaller.
That distinction is the whole risk story. If a sidechain's validators collude or its bridge contract is exploited, base-chain users have no recourse, because the base chain never validated anything that happened there.
Example: a sidechain secured by nine validators holding wrapped assets worth $600m is asking you to trust nine keys with $600m. Several historical bridge losses fit exactly that shape. Size positions there as counterparty exposure, not as on-chain safety.
Related: layer-2, bridge, bridge-hack-risk, wrapped-token