Over the past seven days, the market has been chopping sideways, but Block just delivered a reminder that real adoption doesn't wait for momentum. Cash App and Square drove better-than-expected results, and buried in the shareholder letter was a quieter line: the company has expanded its use of AI across software engineering. I read that sentence twice. Not because I doubt the technology. Because I've spent the last decade building governance systems where every algorithm has a human consequence, and I've learned to listen for the hidden contract inside corporate language. AI is not neutral. The moment a company says it is expanding AI, it is also redefining who is accountable for the output.
Block is not just Square and Cash App. It is the company that put bitcoin on a balance sheet, weathered the collapse of a bull market, and sits at the intersection of regulated finance and open protocols. The results beat expectations as consumer and seller ecosystems grew, and the company has moved past the growth-at-all-costs phase. But the AI line deserves more attention than the earnings beat, because it signals a structural shift in how Block builds software. From my work auditing DAOs, adding an automated layer without human review is how governance breaks. During DeFi summer, I co-designed a quadratic voting system for a community managing over five million dollars. The lesson was simple: technology gives speed, but only a social layer gives trust. That is also why education became my true utility. When I ran Ethical Ledger in Chicago, I taught over 150 retail investors to read smart contracts, and we avoided a fraudulent project that collapsed weeks later. I still believe education is the true utility of blockchain.
Now let's talk about what AI in software engineering actually means. In practice, it means large language models generating code, tests, and documentation, and then human engineers reviewing that output. Done well, this can increase velocity and reduce bugs. Done carelessly, it creates a new dependency layer where the company no longer fully understands its own critical systems. It's not unlike what happened when DAOs began using AI-generated proposals. In my 2026 initiative, Human-First Protocols, I led an audit of AI content in DAO discussions. We reviewed a thousand key proposals and found that the most dangerous ones were not the loudest or the most technically complex; they were the ones that read as perfectly rational but lacked the moral context that only human judgment can provide. The same principle applies to payment software. A code generator can write a function that routes funds, but it cannot decide when a pattern of transactions looks like vulnerable customers being taken advantage of. That requires empathy, and empathy cannot be pretrained out of a model.
If Block treats AI as a way to replace the human layer, it will eventually face a reckoning. But if it treats AI as a way to amplify human judgment, it could build the most resilient fintech stack we have ever seen. The key is what I call human-in-the-loop architecture. That phrase is not a token; it is an engineering decision. It means every AI-generated change gets logged, attributed, and explainable. It means there is a named person responsible for each machine recommendation. In my work with the Values First coalition, I negotiated with a major institution for a ten million dollar grant, and we conditioned that grant on transparency protocols. The executives pushed back, calling our requirements operationally heavy. Our answer was simple: if the algorithm cannot be explained, it should not be trusted. The real innovation in payments is not the model; it is the accountability layer around the model.
What does this mean for Cash App specifically? Cash App has always been the consumer bridge to bitcoin. Over the years, it has weathered regulatory storms and emerged as the most human-centered way to buy small amounts of crypto. The engineering AI expansion could accelerate features like recurring buys, custom spending visualizations, and even a deeper integration with bitcoin lightning. But all of those features require a user's trust. When I audit code, I do not look for cleverness; I look for whether the failure modes are honest. A codebase that has been written mostly by machines hides its failure modes in plain sight. The tests pass, the metrics look green, and then a rounding error steals the equivalent of a month's groceries from someone who trusted the app. That is not a hypothetical concern. Tether has dominated the stablecoin market for years without a truly independent audit, and the whole industry has pretended that problem does not exist. The lesson is that market dominance and technical confidence do not equal trust.
We have seen this pattern in DeFi governance. Vote participation stays below five percent; AI makes it worse. When you expand AI across engineering, you create a governance system for code, and if only machine-output reviewers participate, decision-making concentrates. My advice: publish an AI impact register naming every system where AI touches money and the human responsible for each change. It is not regulation; it is a standard for trust.
Let me be precise about what I think Block is doing right. The company has been more transparent than most banks about its bitcoin holdings and its product roadmap. The move to expand AI inside engineering can be read as a commitment to staying lean while competing with institutions that have massive engineering armies. That is actually a pro-decentralization argument. Open-source AI models and small, accountable teams can unseat legacy gatekeepers, provided they do not sacrifice explainability. But to make that work, Block has to resist the temptation to measure success by the number of AI-generated pull requests. The metric that matters is the mean time to catch a machine's mistake, the number of edge cases a human flagged before they went to production, and the clarity of the explanation when something goes wrong. In a sideways market, these quiet engineering habits are the real signal for long-term positioning. When the next bull cycle comes, the companies that survived will not be the ones with the loudest narratives, but the ones whose software protected the most vulnerable users. Code without compassion is cold, and cold code always fails under pressure.
The contrarian view says this concern is overblown. AI might be a defense against regulatory centralization: without it, only largest banks can afford compliance overhead. So Block's AI expansion could be a survival tool, not a threat. But the blind spot is the review process. If engineers are incentivized to ship faster, the human layer becomes a rubber stamp. In our DAO audits, people who caught problems were rewarded for skepticism. Trust comes from culture, not tech stack, and culture is invisible in a shareholder letter. I'm watching leadership moves more than model choices. If Block pairs AI with clear accountability, it sets a standard. If not, earnings beat marks the beginning of slow erosion.
The bottom line is simple. Cash App and Square beating expectations is good news, but it is not the story. The story is whether Block can integrate AI without becoming the very kind of opaque institution that decentralization was meant to replace. I have spent my career translating code into human consequences, and I believe the firms that win the next decade will be the firms that treat trust as a first-class engineering requirement. That means publishing AI impact registers, rewarding skepticism, and keeping a human name attached to every machine decision. Community resilience is the ultimate hedge. BlackRock and the ETFs have already arrived; the question is whether the human layer still survives the automation. I, for one, refuse to let it disappear. The technology can change, but our compassion cannot. That is why I keep writing. And I will still keep pushing for algorithms with a heart, for leaders who understand that code without compassion is cold.