TL;DR: Bitcoin mining infrastructure firm Luxor Technology says an AI-enabled attack hit its Commander fleet-management software over the weekend, briefly redirecting hashrate from more than a dozen client accounts to a different mining pool. Luxor’s on-call engineering team patched the vulnerability within two hours, says no client data was taken, and plans to fully compensate every account that was affected.

Luxor Technology, the bitcoin mining infrastructure company behind the Luxor pool and the Commander fleet-management platform, disclosed the incident in a series of posts on X. The company said it detected unauthorized redirects of hashrate moving through Commander on Saturday. Its on-call engineering team isolated and patched the vulnerability within two hours of discovering it, then began restoring the affected machines to their original pool settings.
Commander is the dashboard that mining operators use to monitor rigs, adjust settings, and point hashing power at whichever pool they want to mine with. That last function is exactly what the attacker appears to have abused. Instead of trying to drain a wallet or exploit a smart contract, whoever carried out the attack simply rewrote where a batch of client machines were pointed, so their computing power, and the block rewards it would have earned, flowed to a pool the attacker controlled rather than the client’s own account.
Luxor said the incident touched more than a dozen client accounts but that no data was leaked or exfiltrated, and that the problem was isolated entirely to Commander. Every other service, including the core mining pool, kept running normally throughout. The company says Commander is now operating as expected and that affected users do not need to take any further action on their end. It has also committed to fully compensating every client for the hashrate that was diverted during the attack window.

Luxor CEO Nick Hansen addressed the broader pattern behind the attack in his own post on X. “At this moment, anyone running any sort of substantial business, especially in crypto, is inundated with AI-enabled attacks,” Hansen said. “We have taken a very proactive approach. We acquired our own baremetal GPUs to run our own abliterated models and have made huge strides in our security posture.”
That framing matters because it puts a specific label on what is becoming a familiar story across the industry: automated, AI-assisted probing of production software looking for exactly the kind of narrow misconfiguration that Commander apparently had. A fleet-management dashboard that can redirect hashing power is a juicy target precisely because it skips the usual friction of a crypto heist. There is no private key to steal and no multisig to social-engineer. Whoever has write access to where a rig points its hashrate effectively controls where that rig’s earnings go, and an AI-driven scan that finds a hole in that access layer can turn it into free mining capacity within minutes.
The timing is notable for Luxor specifically. The company raised $16 million last month in a strategic funding round led by CMT Digital, with participation from Canaan, MicroBT, Halo, and Steelhead, among others. Luxor’s business already spans a mining pool, ASIC firmware, rig-management software, hardware trading, and energy services, and in September it announced Luxor AI, a push to extend its mining infrastructure into AI and high-performance computing. A security incident on its flagship management tool just weeks after a funding round built partly on confidence in that same software stack is the kind of timing that invites extra scrutiny, even with a two-hour patch and a compensation commitment on the record.
It also lands in the middle of a broader run of security firms flagging just how much automated traffic is now hitting infrastructure that used to only deal with human operators. A HUMAN Security report covered here recently found that agents built on Meta and OpenAI’s platforms now account for almost 90% of all AI agent traffic that security teams are seeing hit their systems, a volume that makes constant automated probing the default rather than the exception for anyone running software with real money behind it. Crypto’s mining and trading infrastructure keeps turning up as a favorite target for exactly that kind of automated attention, right alongside more old-fashioned exploits, like the traders who reportedly made $6 million by watching a public GitHub pull request for clues about an upcoming memecoin listing rather than breaking into anything at all.
Luxor has not published a full technical postmortem yet beyond the X thread, and The Block, which first reported the incident, said it had reached out to the company for further comment. Whether the attacker found the Commander flaw through automated scanning, a leaked credential, or something else specific to Luxor’s deployment is still unclear from what has been disclosed so far.
The bottom line
A two-hour patch window and a public promise to make every client whole is close to the best-case version of this kind of incident, and Luxor deserves credit for saying so plainly rather than burying it. But the attack itself is a useful reminder of where crypto’s weak points are actually shifting. Wallets and smart contracts get most of the attention, yet a fleet-management dashboard that quietly decides where a mining rig’s output goes is just as valuable a target, and arguably an easier one, since redirecting hashrate looks a lot less dramatic on a dashboard than a drained wallet does. As AI-assisted scanning makes it cheaper to find that kind of narrow misconfiguration, the operators who come out ahead will be the ones treating their internal management tools with the same paranoia they already apply to private keys.


