Proof-of-stake blockchains face a specific catalogue of attacks distinct from proof-of-work mining attacks. Cardano's Ouroboros consensus family was designed with each of these categories in mind, and the protocol's security proofs explicitly bound the adversary's success probability under formal models of every threat.
The named attack categories most often referenced in PoS research:
- Long-Range Attack: rewriting old history by building a competing chain from far back.
- Nothing-at-Stake: signing every fork at once because doing so costs nothing.
- Grinding Attack: biasing the leader-election randomness in the adversary's favour.
- Stake-Bleeding Attack: slowly draining stake onto a private fork by replaying fees.
- Posterior Corruption: buying or stealing signing keys from former large stakeholders.
- Bribery Attack: paying current or past validators to misbehave.
- Adaptive Corruption: targeting validators only after seeing who has been elected.
- Stake Majority Attack: the proof-of-stake analogue of the 51% attack.
- Sybil Attack: spinning up many fake identities to gain disproportionate influence.
- Eclipse Attack: isolating a node from the honest network so it sees only attacker-controlled state.
The recurring defences across these attacks are stake-weighted influence (creating fake identities does not help), VRF-based hidden leader election (the adversary cannot target who has not yet been announced), key-evolving signatures (old keys cannot retroactively sign), and Ouroboros Genesis's density-based chain selection (bootstrapping nodes can pick the honest chain without external checkpoints).
Explore next
- Proof of StakeA consensus mechanism where validators are selected to create blocks based on the amount of cryptocurrency they hold and stake (commit) to the network.View term
- OuroborosThe family of proof-of-stake consensus protocols that power Cardano, designed with formal security proofs against the known attack catalogue against PoS chains.View term
- Long-Range AttackAn attack that tries to rewrite long stretches of blockchain history by building a competing chain from far back in the past.View term
- Nothing-at-StakeThe risk that proof-of-stake validators sign every competing fork at once because doing so costs them nothing.View term
- Grinding AttackAn attack on the randomness used to pick proof-of-stake block leaders, where the adversary tries to bias the lottery in its own favour.View term
- Stake-Bleeding AttackA long-range attack in which a minority adversary slowly drains stake from the honest chain onto its private chain by copying transactions and collecting their fees.View term
- Posterior CorruptionAn attack where the adversary buys or steals signing keys from people who held large stake in the past but no longer do.View term
- Bribery AttackAn external attacker pays validators to deviate from the protocol, for example by signing conflicting blocks or handing over dormant keys.View term
- Adaptive CorruptionAn adversary that picks which validators to attack on the fly, after seeing who has been elected to produce upcoming blocks.View term
- Stake Majority AttackAn attack where a single party controlling more than half of the staked ada can rewrite recent history, censor transactions, or double-spend.View term
- Sybil AttackAn attack in which one entity creates many fake identities to gain disproportionate influence over a network.View term
- Eclipse AttackA network attack where a node is isolated from honest peers and connected only to attacker-controlled nodes.View term