Boltz, a non-custodial Bitcoin cross-layer swap service, has suspended its swap service until further notice, a move that has knocked out Lightning-related functions across a range of Bitcoin wallets and payment products that rely on its infrastructure.
The company announced on the night of Aug. 3, 2026 that users should not expect a near-term return of the service and that it could not provide a timeline for reopening. Refund APIs and official support channels will remain online. On the question of how the service may continue in the future, Boltz said it does not currently know.
Boltz said the shutdown was not caused by a single isolated bug. Over the past several months, its infrastructure had been subjected to automated and AI-assisted probing, and it had also gone through several exploit incidents that were contained. In recent days, the pace of attacks picked up sharply. After reviewing recent security scan results, the team concluded that keeping the swap service open would be irresponsible.
The company added that it appears to be facing multiple well-resourced attacker groups, but it did not disclose the specific vulnerabilities involved, the identities of the attackers, or the amount of losses. On user safety, Boltz said funds were never at risk and that any losses were borne by the company. Its API remains available to help process refunds.
Why Boltz mattered to Lightning users who may never have heard of it
To understand why the suspension spread so quickly across wallets and merchant tools, it helps to look at the role Boltz plays.
Boltz describes itself as a “non-custodial Bitcoin bridge.” Its service is used to move value between Bitcoin mainnet, the Lightning Network, Liquid, and RootStock, among other networks and scaling systems. Users do not first deposit funds into a platform account. Instead, swaps are tied together with preimages and timelocks.
A typical example is using L-BTC to pay a Lightning invoice. On one side, the user locks L-BTC on Liquid. On the other, Boltz pays BTC over Lightning. If the payment succeeds, Boltz can use the preimage, the secret value revealed when the transaction completes, to claim the locked L-BTC. If the payment fails, the user can recover the funds. The swap either completes as a whole or falls back to the original assets.
In a June 2024 interview with Bitcoin Magazine, Boltz co-founder and CEO Kilian Rausch said the idea for the project dates back to 2018. At the time, he was building a decentralized exchange based on Lightning and found it difficult to maintain channel balances, leading him to design Boltz as a tool for node liquidity management.
Boltz launched on mainnet in April 2019 with swaps between Bitcoin mainnet and the Lightning Network. The project was initially run on a largely hobbyist basis. In 2023, the founding team decided to work on it full-time and incorporated in El Salvador. In May of that year, Boltz launched Liquid Swap and gradually evolved from a channel-liquidity tool into swap infrastructure sitting behind wallets and payment products.
Bitcoin Magazine also noted at the time that Breez was the first Bitcoin wallet to integrate the Boltz API, allowing users to spend Lightning balances directly for on-chain payments inside the wallet. AQUA later used Boltz to offer swaps between Liquid and Lightning. Before this shutdown, Boltz had been operating in the background of multiple products for years, often without end users knowing it was there.
Three swap types, one hidden dependency
According to Boltz documentation, its service centers on three categories of swaps.
The first is the standard Submarine Swap, used to convert on-chain assets into Lightning payments. The name is usually understood as a metaphor for funds “diving” from the base chain into the off-chain Lightning Network.
The second is the Reverse Submarine Swap, which runs in the opposite direction. The user first pays over Lightning, and Boltz then locks the corresponding assets on Bitcoin mainnet or Liquid. When the user claims the on-chain assets, the preimage becomes public and Boltz uses it to settle the Lightning payment. If Boltz fails to lock the assets, the pending Lightning payment is canceled and the funds automatically return to the user.
The third is the Chain Swap, used to convert assets across different on-chain networks, such as swapping BTC and L-BTC between Bitcoin mainnet and Liquid.
The core appeal of Boltz is that it replaces the need to trust a provider with the need to verify transaction conditions. Boltz supplies pricing, liquidity and coordination, but user assets are not held long-term in the way they would be on a centralized exchange. The service provider cannot simply take one side of the funds after receiving them, and it cannot prevent users from recovering assets after a failed swap or once a timelock expires.
Wallets such as AQUA and BULL used this model to wrap complicated cross-network swaps inside ordinary send and receive buttons. Users did not need to run Lightning nodes, open channels or manage inbound liquidity. Open the wallet, scan an invoice, hit send, and the experience looked close to that of a normal Lightning wallet. Boltz functioned as a transfer station hidden behind the interface.
The shutdown exposed a single point of availability
This design did not erase Lightning’s complexity so much as move it away from the user and onto the swap provider. Users no longer had to manage channels themselves, but wallets became dependent on Boltz for pricing, liquidity and execution.
Once Boltz stopped operating, that dependency was exposed immediately. Users still control their own L-BTC and can continue making transfers inside the Liquid network, but Lightning payments and cross-network conversions that require Boltz have stopped working for now. Boltz was not a custodial single point of failure for funds. It had, however, become a single point of failure for feature availability in parts of the ecosystem.
The most direct impact hit cross-layer conversion features in wallets. BULL Wallet, which relies on Boltz for Lightning payments and Liquid-to-Bitcoin conversion, said Lightning sends and receives inside the wallet, along with swaps between Liquid and Bitcoin mainnet, will fail until an alternative is in place. Bull Bitcoin also said it is a member of the Liquid federation and can convert L-BTC to BTC without relying on a third party, meaning user funds will not be trapped on Liquid.
AQUA Wallet said restoring swaps between Liquid and the Lightning Network is its top priority and added that users still have several routes to convert L-BTC into BTC or USDT while retaining control of their funds. The Bitcoin wallet Zeus has paused its own swap instance.
Blockstream App had previously confirmed in March 2026 that atomic swaps between Lightning and Liquid in Blockstream App and the Jade hardware wallet are powered by Boltz. That means the service halt blocks that route as well.
Merchant tools, fundraising and top-ups were also hit
The knock-on effects did not stop with wallets and swap interfaces.
BTCPay Server said the plugin that depends on Boltz is no longer working properly, leaving merchants who use that plugin temporarily unable to accept Lightning payments. Users who did not enable the Boltz plugin are unaffected.
Bitcoin crowdfunding platform Geyser said both donations and project withdrawals are processed through swaps and cannot continue for now, though user funds are not at risk.
Freedomia, a virtual card project, said Lightning and Liquid top-ups routed through Boltz have been affected. Custody project Mynymbox has also suspended related payment functions.
Taken together, the fallout shows that Boltz is no longer just a swap product visited directly by users. It has become infrastructure behind wallets, payment plugins, merchant tools and other Bitcoin applications. Once those products packaged Boltz into a smooth interface, many users had no reason to know that a “Lightning payment” still depended on an external swap provider until that provider went offline.
AI may be an amplifier, not the full explanation
There is no public technical report showing that every attack against Boltz was AI-driven. A narrower reading fits the available facts better: AI acted as a force multiplier. It can help attackers read public code, generate malformed inputs, traverse API states, combine exploitation paths and keep iterating based on responses.
But AI is not the only variable. Services like Boltz expose open interfaces, involve complicated transaction flows and can offer direct economic upside if vulnerabilities are found and exploited. On the defensive side sits a comparatively small team. That mix makes automated attacks especially effective.
Boltz’s GitHub organization page currently lists five public members. In a January 2026 interview, co-founder and CEO Kilian Rausch also said the project is a self-funded five-person team.
Open source brings transparency, verifiability and community review. It also demands security investment that matches an open attack surface.
Other builders cited in the report echoed that view. PayPerQ, a pay-per-query chatbot project, said it had to respond to an exploit roughly every week over the past several months and believes most of them may have been AI-driven. The operator of the peer-to-peer Lightning Telegram bot lnp2pBot said the bot has been under attack since day one, that the team is using AI to review code in response, and that it may shut the service if the volume of attacks grows beyond what it can handle.
What the Boltz incident leaves behind
Bitcoin security expert and Defending Bitcoin author Luke de Wolf described the Boltz shutdown as a major step backward for interoperability across Bitcoin layers. He said he had long recommended this style of wallet, where Liquid acts as the balance layer and Boltz connects the user to Lightning, while also worrying that Boltz could become a single point of failure. What had been a theoretical risk has now become a real outage.
Early Bitcoin developer Peter Todd, in turn, called these products “fake Lightning wallets” and argued that simulating a Lightning payment experience with Liquid plus a backend swap service is a design worth treating with caution.
The incident also leaves a set of harder questions for wallet and infrastructure developers.
One is that non-custodial products need to do more than show that a provider cannot take user funds. They also need to make clear which functions depend on outside infrastructure, which actions remain available if that infrastructure fails, and how users can recover funds independently.
Another is that wallets cannot treat a single swap provider as if it were a public utility that is always online. A more mature architecture would support multiple swap providers, dynamic routing and automatic failover, or let advanced users connect to self-hosted instances. Even that is not just a matter of adding a few API endpoints. It brings questions around liquidity depth, quote quality, privacy leakage and security standards across implementations.
A third is that security competition in the AI era may force small open-source financial teams to redraw their product boundaries. In the past, open code and non-custodial fund handling were often enough to build trust. Going forward, projects may also need to show that they can sustain automated testing, attack-surface management, abnormal traffic isolation and incident response.