Presented by Stripe Stratechery is now presented by Stripe. What that means for me, for you, and for the future of Stratechery content. stratechery.com/2026/present…

I waited, learned and made my first investment since returning as an active investor at Sequoia. Whop is a special company, and I’m excited to help them to create jobs and stimulate the economy.

How safe is your bank account from AI hacking?

Do people think their bank accounts are safe?

EXCLUSIVE: Vishal Garg fought to regain control of @betrmortgage. Now he wants someone else to be CEO. “I’ve never been really good about EQ,” @vishal_better admits. Speaking exclusively to This Week in Fintech's @bayareawriter, Vishal discusses the boardroom battle that led to his removal, why he's shifting his focus to technology and his plans to bring Better to break-even. With the digital mortgage company still reporting losses, can its AI-powered platform deliver a turnaround? Read the full interview on This Week in Fintech.

Speaking to @ChrisBrummerDr at DC Fintech Week, FDIC Chair Travis Hill casts doubt on agentic AI suddenly causing everyone to move money to banks with highest yield. Notes that there are services today that allow that already.

👻 The banking industry is haunted by the ghost of Silicon Valley Bank. I was thinking about SVB a lot as I wrote my deposit series, and was surprised to see it’s still on others’ minds as well. The latest ghost story is a new report from Starling Insights, commissioned by Fed Vice Chair Michelle Bowman, that looks at regulators’ response to build narrative around the bank. 😵‍💫 It can be tempting to summarize Silicon Valley Bank’s failure as regulatory negligence or exceptional mismanagement. It spares the living from needing to think about how assumptions that seem prudent can become catastrophic — and how hard it is for anyone to update their thinking once they’re stuck in it. 🏦 The ghost of SVB persists because the conditions do too. Deposit pricing competition is intense. Unrealized losses on bank securities sit at nearly $330 billion today. Interest rates are once again creeping up. 👩🏻‍💻 My newsletter this week is a ghost story that asks if we’ve learned enough to avoid it happening again. finitynetwork.com/library/6c…

We will soon see what will be, in effect, a perfectly legal distributed denial of service attack on customer service departments authorized by millions of consumers when it becomes widely known that Muse/Dot/etc can make unlimited requests for old refunds, negotiate deals, etc.

CFTC staff analyzing oil trades for insider trading around the U.S.-Iran war were told to retract their referral to the agency's enforcement division | h/t @CGorrivan bloomberg.com/news/features/…

Mark Zuckerberg pretending to enjoy Guinness is my new favourite thing.

We discovered a sophisticated recording device that was hidden inside one of our offices. We used our models to understand the device and where it had come from, identifying that a specific employee had planted it a few days earlier. We referred the matter to the FBI, and we’re grateful to the agents and prosecutors who moved quickly to make an arrest. The FBI’s investigation is ongoing and we will continue to support their work. At this time, we have no indications of this individual inappropriately accessing or exfiltrating model weights, source code, or customer data. The employee was terminated. We take protecting our technology, data, and people seriously, and as a matter of national importance. Our security program is designed to protect against a wide range of threat actors, from insider risks to cybercriminals to nation states, and this case is an example of how AI can help address these sophisticated security threats.

Last week OpenAI launched Dots, its personal agent, but what caught my eye was a fascinating infrastructure decision they made: Many of the other personal agents don’t self-identify as such to websites they visit. They just look like other bots. Unlike those, Dots is clearly announcing itself as an agent. By doing so, it bypasses the need to rely on Fastly or Cloudflare edge data centers for IP proximity to the end user. This makes our job as fraud fighters easier, but we still have to answer the other two questions. Does this agent belong to this customer? Is the agent doing something the customer would hand off? These are arguably much harder questions to answer than if the bot is actually a personal agent. Another design choice I liked by the OpenAI team - Dots stores your cookies in the cloud. This lets you bypass repetitive OTPs and log back into your favorite sites instantly. Crucially, they’ve paired this convenience with clear, accessible privacy controls that let you easily delete your cookies. Getting security, privacy, and ease of use in the right balance will take time when it comes to these agents but it feels like OAI is on the right track. More to come on fraud and fincrime fighting in the age of personal agents!

I don't think I've seen a topic as wildly, incapacitatingly, intellectually deranging as Agentic Commerce. Note that there are no arguments here; it is taken for granted that a) "people" (who?) will want agents to buy "stuff" (what stuff?) for them and b) "they" (who?) won't want to use "some Amazon-supplied agent to do it" (why not? And how do we interpret Amazon's in incremental annualized sales in 2025 from Rufus / Alexa for Shopping then?). This is not disciplined, logical analysis. This is faith; religious fervor. That some new technological capability becomes the dominant consumer form factor is asserted as inevitable without support. But we know what retail platforms' incentives are: to directly manage consumer relationships so as to induce broader monetization, advertising revenue, and deeper integration (see: Amazon and Walmart commentary on the value of advertising to their businesses, Amazon's deterministic identity spine reaching 90% of US households, Walmart's acquisition of Vizio and vibe, etc.). If all of this is stripped away: what actual opportunity exists to compete with Amazon? In other words, if the consumer relationship is competed away, what's the value of merely operating the agent <-> retailer plumbing? Sounds like a commodity with substantial operating costs if the retail platform is still expected to manage fulfillment and logistics, customer service / returns, etc. Or will the agent do this?

How easy should it be to start a neobank? That's the question that Whop (a digital marketplace and fintech infrastructure provider) raised when it promoted a new capability that allows any creator on its platform to launch a neobank in 15 minutes. This is a real thing. Whop's DeFi and stablecoin-centric financial services infrastructure has been configured into a reusable neobank template, which any of its creators can then customize and launch, using the platform's AI assistant. You can, quite literally, one-shot a fully functioning crypto neobank into production. And, disturbingly, there appear to be virtually no controls or guardrails. You can describe your neobank however you want, even if that description is clearly deceptive. And you can set the prices for the various services offered by your neobank (payments, withdrawals, swaps, etc.) at whatever you want and you can change them whenever you want, with no notice. If you have any experience building in TradFi banking or fintech, you will immediately see the problems with this. And that raises a question: How does something like this even make it to production? I'm not entirely sure, but I think a big part of the answer is Whop's use of stablecoins (Tether recently invested $200M in the company) and DeFi as infrastructure. This infrastructure pushes regulated entities like banks away from the center of the product (funds storage, yield generation, etc.) and out to the periphery (on-ramping/off-ramping, card issuance, etc.) From out there, it can be hard to see how the product itself is being configured, priced, and marketed, let alone enforce any regulatory requirements. There will be demand for this type of infrastructure solution. As a default state, fintech developers want fast, cheap, and easy. That used to be the Bancorp. Then it was Evolve & Synapse. Now, maybe, it's this.

I am a stablecoin guy. My gut reaction when first reading this article was: "Oh! Stablecoins can help bring ethical visibility back to the dollar!" That was the incorrect reaction. Stablecoins are not a silver bullet. The Steelman argument as to why stablecoins actually contribute to illicit financial flows is strong. The article directly mentions that A7 (Corporate agent of illicit Russian Trade) uses both USDT and an onchain stablecoin pegged to the Ruble to evade sanctions and fund the Russian Military. However, what if we ask the rather reductionist question: Is an onchain system (stablecoins) or an offchain system (traditional fiat) best suited to support dollar visibility? For me, the clear answer is stablecoins. The article explains that correspondent banking is simply a chain of private ledgers, and the US only sees the links that pass through its own banks. The best weapon that the fiat system has been able to come up with over the last 20 years is... more regulation of correspondent banks... which has not worked. Tighter correspondent-bank regulation has already been tried, and the article itself shows it backfired: billion-dollar fines caused big banks to de-risk, which created the nesting problem as smaller banks routed through intermediaries. Stronger SWIFT messaging rules already exist on paper and still fail in nested chains because each bank has to faithfully pass the data along. Stablecoins are different because the transparency is an inherent property of the technology, not simply a rule that a bank should follow. An on-chain transfer can't be "netted away" into an invisible bulk settlement the way many illicit financial transactions are. The individual transaction is the settlement. This unveils the counterintuitive nature of stablecoins - they may improve dollar visibility, but they also weaken the chokepoint. Using non-custodial wallets, one bad actor can send another bad actor USDT in seconds. You can see that onchain, and you MIGHT even be able to identify the wallets as bad actors, but you can't do much about it. If the government fines BNP Paribas, it works because BNP Paribas needs dollar clearing from the Fed. But the same cannot be said for a wallet-to-wallet transaction of USDT because finality is not dependant on a bank in NYC releasing funds. Issuer-led freezing sounds like a sexy solution, but it's retroactive and relies on private third parties to make game-time decisions. Blacklisted wallets work too, but when parties can spin up an unlimited amount of wallet addresses, they become less effective. So where do we go from here? Clearly, dollar visibility is an issue. As an American, I want the dollar to proliferate, and I want my government to be able to stop bad actors from weaponizing that dollar. (Only I can weaponize the dollar, dammit). Truthfully, I don't have a perfect answer for you, but I do believe that we are on the right path. As more financial institutions begin to interact with stablecoins, they are able to set stricter requirements around which stables and chains they would like to interact with, which puts pressure on competing stablecoins (and stablecoin-adjacent infra) to clean up their act and work on viable solutions with the black hole of dollar gravity - the U.S. government. Would love to speak to the folks on the forefront of this policy and technology. As always, DMs are open.

I need to know wtf is going on in Montana