Bridges are like rope ladders between skyscrapers—useful, but one weak knot and you’re falling. Most of the largest crypto hacks (Ronin, Wormhole, Poly Network) came from exploited cross-chain bridges. The risk comes from:

• Complex smart contracts that handle assets across chains, expanding the attack surface.
• Centralized validators or multisigs that can be compromised.
• Liquidity honeypots—billions locked in one spot, irresistible to attackers.

That’s why many people are shifting toward native interoperability (things like Cosmos IBC or Polkadot’s XCM) or Sui’s eventual native cross-chain features. These reduce the “wrap and hope” model by letting chains communicate more directly.

Personally? For big sums, I’d avoid third-party bridges unless they’ve been heavily audited, battle-tested, and ideally insured. For small amounts, bridges can be fine, but you have to treat them as high-risk.

The interesting frontier now is projects building trust-minimized bridges (using zero-knowledge proofs or light clients) so you don’t rely on a small validator set.