← Back to Blog
Web3 Security · 9 min read · November 18, 2026

Is Restaking Safe? Slashing, Exit Queues and LRT Risk in 2026

Restaking is safe in the narrow sense that nobody has to hack anything: the mechanism does what the documents say, and the loss arrives as a rule you agreed to. What 2026 revealed is that the risk does not announce itself as an exploit. It leaves quietly — the largest liquid restaking protocol pulled restaking out of its flagship token, and the sector’s deposits went with it.

What restaking actually asks you to accept

Restaking reuses ETH that already secures Ethereum to secure extra services, called actively validated services or AVSs, on a shared-security layer such as EigenLayer or Symbiotic. The path is short and every step moves control away: you deposit and delegate, a node operator runs the service, the operator opts into an Operator Set, and it allocates your stake to it. From that moment your ETH is slashable under rules the service wrote.

Two defaults concentrate the exposure: delegation is all or nothing to one operator, everything staked afterwards auto-delegates to the same one, and an EigenPod cannot be transferred.

The distinction that matters is between the two kinds of penalty. Beacon-chain slashing follows Ethereum’s consensus rules, which are published, reviewed and rare. AVS slashing follows whatever the service decided misbehaviour means. You cannot read every AVS’s conditions the way you read a consensus spec: you are trusting an operator’s judgement and an AVS’s governance, not just a protocol.

The five risks stacked in one position

What 2026 actually did to the sector

On 6 August 2026 ether.fi removed all restaking exposure from weETH, turning its flagship asset into a plain liquid staking token. Restaking moved to weETHs, a separate token on Symbiotic holding 9,136 tokens worth about $18 million — roughly 0.5% of the staking base, against 1.72 million weETH in circulation.

Its own risk documentation, updated the same month, put less than 1% of assets as still restaked with EigenLayer, down from about half in early 2026, with a plan to reach zero in Q3 2026 and strip EigenPod withdrawal credentials from its validators by Q4 2026. The CEO’s public comment was close to a shrug: end of an era, and restaking was probably just a bit too early.

The sector numbers follow the same line. EigenCloud sits near $5.10 billion against a $22.06 billion peak in August 2025; Symbiotic near $343 million against $2.70 billion in December 2024. ether.fi’s staking TVL was about $3.8 billion on 17 November 2026, down 24% in thirty days. Its redemption record is instructive too: 542,792 ETH, 19.6% of TVL, in 33 days during a rush for the exits, median claimable time 4.9 days.

None of this was a hack. It was a repricing. The 2024 yields were points and airdrop incentives; when those stopped, revenue had to come from real service fees. A March 2026 econometric study of Renzo found protocol revenue tracked EigenLayer’s TVL, the LRT’s yield and its multi-chain expansion — not the protocol’s deposit base. Yield paid to acquire deposits is a subsidy, and subsidies end.

Seven questions that size an LRT in ten minutes

  1. How much is restaked right now — today, not at launch? The number is usually in the risk docs, and it is the only one that matters.
  2. Which operators, and how concentrated? One operator running most of the stake is a single point of failure with extra steps.
  3. What share sits in redistributable sets? Higher headline yield here is the market pricing the right to take your funds.
  4. Which AVSs are attached, and are their slashing conditions and upgrade controls public? A timelock is not a guarantee, but its absence is a red flag.
  5. Where is the exit? Queue-bound or instant, and how long is the queue this week?
  6. Does the discount exceed the staking token’s? If an LRT trades wider than stETH, the market is pricing something the docs are not telling you.
  7. Do the docs match the contract? In this cycle they frequently did not: one protocol still described its flagship token as auto-restaking months after the restaking was removed.

How to hold the exposure without mistaking it for staking

Treat restaking as a credit position, not a savings account. Size it so a 10% slash and a three-week exit are both survivable, and assume two failures rather than one: slashing and a long queue tend to arrive in the same week. Do not loop the receipt token back into lending; recursion turns a bad week into a liquidation. Read the risk page before the yield page, in that order — the yield number is the one written to be read.

Frequently Asked Questions

Is restaking the same as staking?

No. Staking puts ETH at risk of Ethereum’s own consensus penalties. Restaking adds a second layer: an operator can allocate your stake to services that write their own penalty rules, so the same ETH can be slashable by several parties at once. It also adds operator trust and exit-queue risk that plain staking does not carry.

Can I lose money restaking without being hacked?

Yes, in at least three ways. Your stake can be slashed by an AVS, the liquid restaking token can trade below the value of the ETH it represents, and the yield can be smaller than the cost of waiting in an exit queue during a rush. None of those need an attacker; they are the instrument working as designed.

Why did ether.fi remove restaking from weETH?

The company has not published a detailed explanation. Its risk documentation disclosed that restaked assets fell below 1% of the protocol, down from about half in early 2026, with a target of zero in Q3 2026 and full removal of EigenPod withdrawal credentials by Q4 2026. Publicly it framed the change as giving holders a clearer choice, and its CEO said restaking was probably a bit too early.

What happened in the November 2026 EigenLayer slashing incident?

An operator deployed its own AVS, allocated its own stETH to it and slashed about 90.5% of that stake, then received 31.1 stETH of the slashed funds in its own wallet. ether.fi’s monitoring read the release as a possible exploit and automatically paused ten contracts for 51 minutes. No user funds were affected and no ether.fi stake was slashed, and EigenLayer confirmed the slash was intentional.

Is restaking dead?

Not dead, smaller. Sector deposits topped $18.3 billion in December 2024 and have contracted sharply since the points and airdrop incentives that drove them ended. The model still works where the service it secures earns real fees, such as bridges, oracles and underwriting capacity. What ended is the version where the reward came from the subsidy rather than the service.

Yield you did not model is risk you did not price

The same discipline applies to the protocols you interact with. RootCrak’s autonomous scanner checks domains, front-end assets and cloud surfaces for the tampering and impersonation that precede a loss.

Get a Free Security Scan