The non-custodial bitcoin service Boltz is undergoing a change of leadership. The project's founders have decided to completely step away from management after a series of targeted attacks that, in their assessment, were carried out using artificial intelligence tools. This event marks a new round of escalation in the confrontation between cybersecurity and automated threats.
The nature of the incidents, as the developers note, has shifted toward increasing frequency and complexity. Some of the attacks were successful, leading to direct financial losses for the service's operators. The key point: user funds remained safe, since Boltz's architecture does not involve storing client assets. The full brunt was borne by the developers themselves, which underscores the resilience of the non-custodial model in the face of aggressive actions by malicious actors.
As a reminder, the platform specializes in atomic swaps, facilitating exchanges between the bitcoin mainnet, Lightning Network, and Liquid Network. After operations were suspended on August 3 to prevent further damage, the service has still not been restored. On the Boltz homepage, an official notice has been posted about the transfer of the project to a new team.
Notably, the small team of five people admitted their inability to withstand attacks of this level on an ongoing basis on their own. The new owner will be an undisclosed group of "bitcoin veterans" ready to provide not only capital but also significant engineering resources. The new owners have already begun an audit and vulnerability search, hoping to resume swap operations in the near future. All original founders immediately left their positions, retaining only the option of voluntary participation in the development of the open-source code.
AI as a New Factor of Power in Cyber Warfare
Almost simultaneously with the incident surrounding Boltz, the Bitcoin Policy Institute (BPI) and a number of leading industry players, including Block, Coinbase, Kraken, and Ledger, published an appeal to AI developers. The essence of their request is the creation of controlled trusted access programs to the most powerful models for legitimate security researchers and developers of critical financial infrastructure.
The letter's authors rightly point out an imbalance: AI is already capable of analyzing vast arrays of code and finding vulnerabilities faster than a human. However, defenders of open-source software often lack access to these same tools due to restrictions imposed by the AI services themselves. A paradox arises where attackers obtain more advanced weaponry than those standing guard. BPI proposes not to lift all restrictions, but rather to build verified access channels for trusted teams, which looks like a pragmatic and timely step.
Against this backdrop, the loss statistics look even more alarming: in the first half of 2026, crypto projects lost about $1.1 billion due to hacks, and the number of confirmed exploits reached record levels over the past six months. This case with Boltz is a vivid illustration that AI has already become a real weapon in the hands of malicious actors, and the industry urgently needs to adapt to the new realities, otherwise we risk seeing even larger-scale incidents.