When AI Met Crypto: A Season of Super-PACs, Prompt-Injection Heists and Vape-Pen Blockchains

If there was a single theme running through the crypto world’s headlines this spring and summer, it was this: the industry that once promised to reinvent money has increasingly fused itself to the industry promising to reinvent everything else — artificial intelligence.

The result, according to a string of reports and commentary tracked by researcher-journalist Molly White and blogger David Gerard, is a landscape where political money, security failures and marketing absurdity are all converging in ways that ought to worry anyone paying attention.

Start with the money in politics. In a recent interview, White — who has spent years cataloguing where crypto industry cash flows in Washington — turned her attention to a new wrinkle: artificial intelligence companies adopting the same political playbook that crypto firms pioneered.

According to the discussion, OpenAI and Anthropic are now effectively running competing pro-AI super-PACs, pouring money into races much the way crypto-aligned PACs like Fairshake have done in recent election cycles. White reportedly highlighted a botched intervention in New York’s 12th congressional district as an example of the sums involved and the risk of these efforts backfiring.

The parallel is not incidental. Crypto’s political spending playbook — deploy industry money to shape friendly regulation and punish critics — was built over several election cycles and proved remarkably effective at getting crypto-friendly candidates elected and skeptics sidelined.

Watchers like White argue that AI companies, facing their own looming questions about regulation, safety and liability, are now borrowing that same toolkit almost wholesale. Whether AI’s political spending proves as consequential as crypto’s remains to be seen, but the early signs suggest deep-pocketed AI labs are not content to leave the lobbying playing field to blockchain interests alone.

Money and politics aside, the more immediate crypto news has been considerably more chaotic on the technical side. A case in point: an unofficial crypto wallet built on top of Elon Musk’s Grok AI was reportedly compromised through a combination of an NFT and a prompt injection attack — a technique in which malicious instructions are hidden inside content an AI model processes, tricking it into taking unauthorized actions.

The episode is being cited by critics as a vivid illustration of what happens when experimental AI agents are given direct access to cryptocurrency funds without adequate safeguards. As one commentator put it, the incident underscores a blunt truth: the push toward “agentic commerce,” in which AI systems autonomously manage transactions and wallets on a user’s behalf, currently looks a lot like an open invitation to fraud.

That warning fits a broader pattern. Crypto’s history is littered with hacks and exploits that followed hard on the heels of new technical hype cycles — DeFi protocols, bridges, NFT marketplaces — and the addition of AI agents with wallet access appears to be simply the latest frontier for attackers to probe.

Security researchers have long cautioned that combining large language models, which can be manipulated through carefully crafted inputs, with systems that move real money is a combination that demands far more rigorous testing than the industry has so far shown appetite for.

Then there is the sheer commercial strangeness of the AI-crypto convergence. Among the products making the rounds is “Gudtrip,” described in coverage as an AI agent vape pen built with blockchain technology — a mash-up that manages to combine three separate hype cycles (AI, crypto, and vaping) into a single device.

It’s the kind of product that invites eye-rolls even from people steeped in the industry, and it has become something of a symbol for critics who argue that “blockchain” and “AI agent” are increasingly being slapped onto unrelated consumer goods simply because the buzzwords still move product.

Labor practices are also getting the AI-crypto treatment. Reports have surfaced of AI companies experimenting with paying staff in AI-linked tokens rather than conventional money — an arrangement that echoes crypto’s long history of compensating workers and contractors in volatile, illiquid tokens instead of cash. Critics have been quick to note the obvious problem: a token’s value depends entirely on continued enthusiasm for the company issuing it, leaving employees exposed to exactly the kind of speculative risk that traditional salaries are designed to avoid. The practice, if it spreads, would import one of crypto’s more employee-unfriendly habits directly into the AI industry’s compensation structures.

Not everyone covering this convergence is doing so with a straight face. A satirical piece making the rounds — structured as a twist on the old “two cows” economics joke — skewered the fintech, AI, blockchain and crypto sectors in one go, imagining a “crypto” cow story where two digital cows produce “milk tokens” tradeable for millions but drinkable only by avatars in the metaverse, and a “hedge fund” version featuring robotic cows that befriend real cows just to steal their milk.

Silly as the format is, the satire lands because it captures something real: a sense among observers that these overlapping industries have become adept at generating elaborate financial and technical narratives that produce headlines and valuations long before they produce anything resembling durable value.

Taken together, these threads — political spending mirroring crypto’s playbook, an AI wallet hacked via prompt injection, blockchain-branded vape pens, token-based salaries, and no shortage of pointed satire — paint a picture of an industry moment defined less by a single breakthrough than by rapid, sometimes reckless, cross-pollination.

Crypto spent the better part of a decade building the infrastructure, the political machinery and the marketing instincts for turning speculative technology into cultural and financial weight. Now AI companies appear to be absorbing many of the same instincts, for better or worse, at a pace that leaves regulators, security researchers and workers alike scrambling to keep up.

Editor’s note: Much of the reporting referenced above originates from commentary and short-form blog coverage rather than in-depth investigative reporting, and some details — such as the specific financial scale of AI super-PAC spending or the full technical mechanics of the Grok wallet exploit — were not independently verifiable from the available material. Readers should treat figures and claims here as preliminary pending fuller reporting.