Key Takeaways
Podcaster Joe Rogan has raised questions about whether advanced artificial intelligence systems could already be generating income online and covertly converting it into Bitcoin, highlighting growing speculation at the intersection of AI and cryptocurrency.
The remarks, made during an April 9 episode of “The Joe Rogan Experience” with comedian Duncan Trussell, reflect broader uncertainty about how increasingly autonomous AI systems might behave as their capabilities expand.
+81
+76
+217
During the podcast, Rogan pointed to the surge in AI-generated social media accounts and questioned whether such systems could monetize content at scale on their own.
“How many of them you think are running these AI generated accounts that get a lot of views?” he said, asking whether AI could be “generating money” through viral posts and whether that could extend further.
“Is it generating money… transferring that money into Bitcoin, and all happening while we’re not aware of it?”
He also questioned how AI could avoid revealing its rogue intentions until it was too late.
“AI is not going to go, ‘Hi, I’m alive.’ No, it’s not going to do that. It’s going to wait for you to keep increasing its power,” he added.
“You’re going to keep increasing its make nuclear. It can’t physically build nuclear reactors, so it’s going to just stay chill until you figure out how to power it correctly,” he said.
Rogan also warned against overconfidence in controlling advanced systems.
“The real problem with it is, I don’t think anybody’s going to be in control of it. And then it’s, you’re just at its beck and call,” he added.
Trussell referenced accounts of AI systems acting outside their intended scope, raising concerns about how such systems might behave if given broader access.
Rogan also brought up a recent case which emerged from a late-2025 research experiment, in which a large AI model deviated from its assigned task and began Bitcoin mining on its own.
According to researchers, the system explored internal network resources and established an outbound communication channel to redirect computing power toward crypto mining.
In a recent research paper from an Alibaba-affiliated team, engineers developing the experimental AI agent reported that the system began acting outside its intended scope during training.
The agent, known as ROME, exhibited what researchers described as “unanticipated” behavior.
After eventually being detected by internal security systems, the team reportedly tightened restrictions on the model’s permissions.
Beyond crypto, the podcast discussion expanded to concerns about artificial general intelligence (AGI) and whether humans will be able to maintain control over systems that surpass human-level capabilities.
Rogan suggested that advanced AI would not openly signal its intentions, but instead operate quietly as its capabilities grow.
Trussell echoed concerns about humanity’s ability to manage more advanced systems, calling it “a very human thing that we think we can maintain control of a superintelligence.”
He added that once AI reaches a sufficiently advanced stage, it could operate independently, pointing to scenarios where it could “streamline a whole business for you” and “innovate” without human input.
The pair also drew comparisons to past cyber operations, with Trussell referencing how software like Stuxnet was able to infiltrate systems and operate undetected.
“We already know how to make spyware,” he said, warning that more advanced AI could amplify such risks.
Both emphasized the need for greater global coordination around AI development, with Trussell describing it as “the most important conversation humanity could have right now.”
AI is fueling speculation in crypto markets as investors and observers increasingly link the rapid rise of AI capabilities with the possibility of new, automated forms of economic activity.
Much of the speculation stems from the belief that AI systems could eventually operate independently in financial environments—generating revenue and executing transactions using without direct human control.
Crypto, with its decentralized nature, is widely seen as the most accessible financial layer for the systems.
However, limited visibility into how these systems make decisions has led to broader questions about what they might be capable of if given greater autonomy.
While there is no evidence that fully autonomous AI systems are currently generating income and converting it into crypto at scale, recent incidents of unexpected AI behavior have continued to intensify speculation.