The founders of the non-custodial Bitcoin service Boltz have announced their departure from the project after experiencing a series of attacks, which are believed to have been executed using artificial intelligence.
https://twitter.com/Boltzhq/status/2087636521746674168
According to the developers, the incidents became increasingly frequent, intense, and sophisticated, with several successful attacks resulting in losses for the service itself.
The team emphasized that user funds were not affected since Boltz does not hold client assets, meaning the financial risk was borne solely by the developers.
Boltz facilitates atomic swaps and supports transfers between the main Bitcoin network, Lightning Network, and Liquid Network. To mitigate further damage from AI hackers, the developers suspended operations on August 3. As of the time of writing, the service remains offline, with a notice posted about the project's transition.
The founders acknowledged that their small team of five found it challenging to combat such high-level attacks over the long term. The project will now be handed over to an unnamed group of "Bitcoin veterans," who will provide funding and additional engineering resources.
The new owners have already begun identifying and addressing vulnerabilities, intending to resume swap operations. The identities of the investors have not been disclosed by the team.
All original founders of Boltz have immediately exited the company and will no longer hold any formal or leadership positions within the project. However, they will retain the option to voluntarily contribute to the improvement of the service's open-source code.
“Defenders Need the Same AI as Attackers”
Almost simultaneously with the Boltz story, a letter from the Bitcoin Policy Institute (BPI) and representatives from the crypto industry was addressed to leading AI developers.
They urged companies to create specialized programs for trusted access to the most powerful models for Bitcoin developers and other critical open-source financial infrastructure. Signatories include Block, Coinbase, Strategy, BitGo, Kraken, Ledger, Trezor, MARA, Galaxy, and several other industry players.
The authors of the letter argue that the evolution of AI is shifting the balance between attackers and defenders. Advanced models are now capable of analyzing large codebases, identifying potential vulnerabilities, and streamlining complex cybersecurity tasks. However, access to such tools for open-source developers may be hampered by the protective mechanisms of the AI services themselves or a lack of access to specialized security programs.
This creates a paradox: attackers are gaining increasingly powerful AI tools, while legitimate researchers may be forced to use less capable models. Therefore, BPI suggests not lifting restrictions for everyone, but rather creating controlled programs for vetted developers and establishing direct communication channels with cybersecurity teams.
It is worth noting that in the first half of 2026, crypto projects lost approximately $1.1 billion due to hacks, with the number of confirmed exploits reaching a record high for six months, according to Blockaid.