Mark Cuban has a new hobby. The billionaire entrepreneur, Shark Tank fixture, and former owner of the Dallas Mavericks is running a Mac Mini M4 Pro around the clock in his home, using it to filter out the overwhelming flood of AI-generated emails that have begun clogging his inbox. It’s a small machine solving what Cuban sees as an enormous problem — one that’s only getting worse as generative AI tools make it trivially easy to produce convincing, personalized cold outreach at industrial scale.
The setup, which Cuban described in detail to Business Insider, is deceptively simple. He runs a local large language model on the compact Apple desktop — a machine that starts at around $800 — and feeds his incoming emails through it. The model scores each message for the likelihood that it was generated by AI, flagging or filtering those that cross a threshold. No cloud service required. No monthly subscription. Just a small silver box humming away on a desk.
“The amount of AI-generated emails I get is insane,” Cuban told Business Insider. He said the volume has increased dramatically over the past year as AI-powered sales tools have proliferated. What used to be a manageable trickle of cold pitches has become a torrent of polished, AI-crafted messages — each one designed to look personal, each one consuming a few seconds of attention before the recipient realizes it’s junk.
Cuban isn’t alone in noticing the shift. Across industries, executives and investors report that their inboxes have become war zones. Venture capitalists say AI-written pitch emails now outnumber human-written ones. Sales leaders describe an arms race in which every team uses AI to send more outreach, which means every recipient gets buried in more noise. The math is brutal: if AI lets one salesperson send ten times as many emails, and every salesperson adopts the tool, inboxes don’t just get busier. They become unusable.
The irony is thick. AI created the problem. And now Cuban is using AI to fight it.
His approach reflects a broader trend among technically inclined professionals who’ve begun deploying local AI models for personal productivity tasks. Running a model locally — rather than sending data to OpenAI, Google, or Anthropic’s servers — offers privacy advantages and eliminates per-query costs. Apple’s M-series chips, with their unified memory architecture, have turned consumer-grade hardware into surprisingly capable inference machines. The Mac Mini M4 Pro, with its 24GB of unified memory in the mid-tier configuration, can run quantized versions of open-source models like Meta’s Llama 3 or Mistral’s offerings with reasonable speed.
Cuban has long been an early adopter who relishes tinkering. He made his first fortune selling broadcast streaming company Broadcast.com to Yahoo in 1999. More recently, he launched Cost Plus Drugs, an online pharmacy that publishes its markup transparently, and he’s been vocal about AI’s potential across healthcare, media, and business operations. But his email-filtering project isn’t a startup pitch. It’s a personal fix for a personal annoyance — the kind of practical, unglamorous application that often signals where technology is actually headed.
The spam problem he’s addressing has deep roots. Traditional spam filters, built into Gmail, Outlook, and other email clients, were designed to catch Nigerian prince scams and pharmaceutical ads. They look for known spam signatures, suspicious links, and sender reputation signals. They work well against bulk junk mail. But AI-generated cold outreach is different. These messages are grammatically flawless. They reference the recipient’s company, job title, and recent public statements. They mimic the tone of a real human who’s done real research. Conventional filters often let them sail right through.
That’s precisely why Cuban’s local LLM approach has an edge. A language model can detect patterns that rule-based filters miss: the slightly too-perfect sentence structure, the formulaic personalization, the uncanny smoothness that marks machine-generated prose. It’s fighting fire with fire — or more accurately, fighting pattern generation with pattern recognition.
The concept has attracted attention from others in tech. Several startups are now building AI-powered email triage tools, though most rely on cloud-based models. SaneBox, which has offered AI email filtering for years, has reported increased demand. And Google has begun integrating Gemini-based features into Gmail that can summarize and prioritize messages, though these don’t specifically target AI-generated outreach.
But Cuban’s DIY approach carries a philosophical charge that these commercial solutions don’t. By running everything locally, he keeps his email content on his own hardware. No third-party server sees his messages. For a public figure who receives sensitive business proposals, legal correspondence, and personal communications, that’s not a trivial consideration. It’s a meaningful privacy stance.
The broader implications extend well beyond one billionaire’s inbox. The explosion of AI-generated outreach is already degrading the signal-to-noise ratio of email as a communication medium. Some investors have publicly said they no longer read cold emails at all. Others have moved critical communications to Signal, iMessage, or private Slack channels — essentially retreating to gated communities within the digital world. If email becomes synonymous with AI slop, the medium itself could lose its utility for professional communication.
Sales teams are feeling the backlash too. Response rates to cold email have plummeted industry-wide, according to data from outreach platforms. The very tools that promised to supercharge sales pipelines are poisoning the well. When everyone can send a thousand personalized emails a day, nobody’s emails feel personalized anymore.
Cuban has been characteristically blunt about this dynamic. He’s argued on social media that AI-powered sales outreach is approaching a point of self-destruction, where the tools become so effective at generating volume that they destroy the channel’s effectiveness entirely. His Mac Mini project is, in a sense, a one-man proof of concept for the countermeasures that will inevitably follow.
And countermeasures will follow. The history of internet communication is a history of spam and anti-spam technology evolving in lockstep. Email spam begat Bayesian filters. Comment spam begat CAPTCHAs. Robocalls begat call-screening apps. AI-generated outreach will beget AI-powered filtering. The question is whether the filtering tools will be built into platforms like Gmail and Outlook or whether users will have to cobble together their own solutions, as Cuban has done.
For now, the Mac Mini sits in Cuban’s home, quietly reading his email, sorting the real from the synthetic. It costs him nothing beyond the initial hardware purchase and a bit of electricity. No API fees. No monthly charges. Just a local model doing what local models do best — processing private data without sending it anywhere.
It’s a small, practical, slightly nerdy solution. Which is exactly why it matters. The most telling technology applications aren’t the ones announced at keynotes with dramatic lighting. They’re the ones that solve annoying problems for busy people. Mark Cuban is a busy person. And his inbox, like yours, is under siege.
The $800 box on his desk might be the first shot in a much larger war.
WebProNews is an iEntry Publication