The Warsh Fed, ChatGPT in your finances, and the Everlane obituary

Five concentrations this week: the Fed, ChatGPT, the MBA, The Voting Rights ACT, and Everlane. If you are not reading the rest, here is what to leave with.

TLDR: Kevin Warsh has been confirmed to run the Fed and your mortgage rate is not dropping. Move idle cash into a money-market fund earning close to 5% while you can. ChatGPT can see your bank account as of Friday. The product is good for budgeting and scenario math, but pair it with Perplexity for research and read the privacy fine print before you connect anything. Mid-tier MBAs are 50% off because the credential stopped paying, and 57% of American workers are afraid to job-hop. Everlane is reportedly selling to Shein for $100 million, which is what happens when you choose money > mission. The Supreme Court weakened the Voting Rights Act in three rulings this week, reshaping the 2026 midterm map.

The Fed has a new chair. Your mortgage rate is not coming down.

Kevin Warsh was confirmed as the next Fed chair last Tuesday in a 54-45 Senate vote, the narrowest in modern history. He is expected to be sworn in on Friday and to follow with faster interest rate cuts than Powell. He’s also called publicly for "regime change" at the Fed. He means three things:

He wants the Fed to speak less publicly. Warsh believes public signaling boxes the Fed in when the data (or their interpretation) changes.

He wants to switch the inflation gauge. The Fed today watches a measure that strips out food and energy. Warsh wants a broader one that captures more real-world prices, including a number he has floated called a "survey of a billion prices." 

And he wants to pull the Fed back from areas like climate risk and bank supervision, on the view that the Fed has wandered too far from its core job of setting interest rates.

In a recent op-ed, he also wrote that AI "will be a significant disinflationary force." That is the cover he is laying down for cutting rates faster than the inflation data alone would justify.

But long-term interest rates went up all week, in the U.S. and abroad. 

The U.S. government borrows by selling bonds. Long-dated bonds (the 10-year and 30-year Treasury notes) pay a fixed interest rate, called the "yield," until they mature. When yields rise, it means investors are demanding more interest to lend the government money for that long, usually because they are worried that inflation will erode their returns over time. 

Oil benchmarks are also still around or above $100. The Strait of Hormuz, where roughly 20% of the world's oil moves through, is still effectively shut. The IEA (International Energy Agency) warned Monday that global oil inventories are depleting rapidly. 

Expensive oil keeps inflation higher for longer. Higher inflation for longer makes long-term bond investors demand more interest. That is how the war ends up in your 30-year mortgage rate.

The G-7 meeting was another political tell. Voters in every G-7 country want their governments to spend more, not less. Bond buyers are increasingly saying: we will not lend the money to do that without higher rates. Higher rates are the market's way of telling these governments to spend less, raise taxes, or both.

There is a version of next year in which Warsh cuts rates, inflation gets worse, the bond market sells off harder, and the gap between what the Fed controls and what the market controls becomes the story. There is also a version where the war ends, oil drops, inflation cools, and Warsh looks like a hero. The bond market right now is betting on the first version.

So for now: capitalize on the fact that short-duration money-market funds are paying close to 5%, the highest in twenty years.

ChatGPT moved into your bank account. Don't default to it for everything.

OpenAI launched a personal finance product for ChatGPT Pro users on Friday, May 15. A federal jury in California tossed Elon Musk's lawsuit against the company on Monday, May 18, the one that could have forced OpenAI and Microsoft to pay up to $150 billion into the company's nonprofit foundation.

Starting with the product. You can connect your bank, brokerage, and credit cards to ChatGPT through Plaid, the same infrastructure powering Venmo and Robinhood. ChatGPT can see balances, transactions, investments, and liabilities. It can’t see full account numbers or take actions like placing trades. Intuit support is coming, which means TurboTax and the rest of the personal-finance stack are next.

What it does well, per early users: categorizes spending in real time, surfaces the fee load on your investment accounts, and runs scenario math on tax decisions like Roth conversions and deduction bunching. The bar for success is accomplishing what a private banker would do for you if you had eight figures to justify hiring one. People who have used it say it is meaningfully better than the chatbot version because it can look at what you actually spent. 

Before you default to ChatGPT for everything, know what Perplexity does that ChatGPT does not. Perplexity cites every source on every answer and surfaces conflicting sources by default. Perplexity Finance is a dedicated finance surface with earnings data, market data, SEC filings, transcripts, and source-linked research, making it better suited for research than a connected-account budgeting assistant. The right setup for most people: ChatGPT for inside-your-life agent work and Perplexity for research. 

Warnings: OpenAI says financial conversations follow each user's ChatGPT model-training settings. But financial information in your conversation history has to be deleted by you. When The Record asked experts about the privacy implications, they flagged obvious risks: anyone with access to your ChatGPT login can see your balances, transaction history, and debt profile. 

Now the lawsuit. In the Northern District of California, Judge Yvonne Gonzalez Rogers accepted an advisory jury's recommendation that Musk took too long to file. The case never reached the merits. Musk called it a "calendar technicality" on X and said he would appeal. Statute of limitations is a real argument, not a technicality, but if you were Musk, technicality is how you’d frame it.

OpenAI won this round without the court reaching the founding-promise question of whether the company violated its nonprofit commitments. That fight is still legally available to someone else. For now, OpenAI has two things at once: the biggest consumer product in AI sitting inside your bank account, and a federal ruling clearing the path for whatever it wants to do next with its corporate structure. 

MBAs at 50% off, Meta cuts 8,000, and your job is the safest place to plan your exit.

American business schools like Purdue, Johns Hopkins, and UC Irvine are running discounts of up to 50%. Applications are down 20% to 30% and in some cases even more than 40%. 

The schools cutting costs are mid-tier and below, which frankly may never have been a worthwhile investment for those paying out of pocket. The top 20 aren’t cutting. The MBA is doing what positional goods do when demand softens: the top holds the line, the rest discounts.

The cause is the thing every working person feels. AI is making midlevel knowledge work feel less safe. The two-year, $200,000 bet only worked when post-MBA jobs were stable and the pay bump was real, but there is far less guaranteed when consulting firms are shrinking junior classes, banks are automating analyst work, and corporate strategy teams are smaller. 

These numbers are caused by what ResumeBuilder calls job hugging. Fifty-seven percent of American workers identified as job huggers in February 2026, up from 45% in August 2025. 

Evidence that workers aren’t crazy shows in the fact that, despite extraordinary profitability, Meta begins laying off 8,000 more people tomorrow. The company is also canceling 6,000 open roles, making the effective reduction closer to 14,000. Cumulative cuts since 2022 are now over 33,000. Tech is up to 113,000 cuts year-to-date.

The half-life of job hugging is shorter than it feels. The real path through is building an earning line your employer does not control. It is harder than it sounds. The math is getting clearer by the quarter.

Everlane is selling to Shein for $100 million.

Everlane is reportedly being acquired by Shein for about $100 million, per Puck News, which broke the story on Sunday. Everlane is carrying roughly $90 million in debt. The equity value of the Everlane brand in this transaction is approximately $10 million. Common shareholders will not receive a payout. Shein is mostly buying out Everlane’s creditors.

Everlane has completely failed. 

Everlane was founded in 2011 by Michael Preysman and Jesse Farmer on a single thesis: radical transparency. The original promise was that you could see what every garment cost to make. Materials, labor, transport, duties, markup. The factories were rated for working conditions and the materials were sourced for traceability. A generation of professional women dressed for their first big jobs in Everlane silk shirts and Day Heel pumps because the brand sold the idea that capitalism could be improved from inside, if customers paid a little more for clothes made fairly. 

Shein is the opposite. Yale researchers have called it one of the industry's most damaging polluters. It has faced credible accusations of forced labor in its supply chain. Its garments are not designed to last. Its prices work because someone else is paying the difference. Garment workers, the environment, and the long-term resale value of the clothes themselves.

Selling Everlane to Shein is the maximum possible inversion of the company's stated reason for existing. Everlane’s early page about why transparency mattered used to argue that mainstream fashion's opacity was the lever that made exploitation possible. And now they are selling out to one of the worst in the business. 

Why is Shein even buying it? Shein has been trying to IPO for years, but keeps getting blocked. It's 2023 US filing stalled after lawmakers from both parties raised concerns about cotton sourced from Xinjiang and accusations of forced labor in its factories. Shein’s pivot to list in London hit the same wall. To convince regulators and big investors that it is a company worth backing, Shein needs a story about ethics and sustainability, one that its own factories cannot provide. A company like Everlane, with B Corp status and a transparency mission already built in, can. Shein is not buying Everlane for the customers or the products. It is buying its reputation. 

How did Everlane end up here? Radical transparency was a marketing position. The business model underneath was a premium price, an ethical halo, and growth at rates outside capital demands. Those are not compatible inputs. Sustainable apparel margins cannot grow at venture pace; ethical supply chains cannot scale at private-equity pace. Three CEOs in five years, prices crept up while perceived quality did not, transparency pages got harder to find, and factory lists got vaguer. Almost every honest, ethical fashion brand that took growth capital has ended in a sale to someone less ethical. Everlane just took the most extreme version of that ending available.

There are real brands doing this real work and you can find them. The narrower lesson is that ethics built on top of growth capital is provisional. Many brands that build their identity on values and accept outside money are one bad quarter away from selling the values back. Patagonia is what an honest version of this looks like: private ownership, refusal of growth capital, and founder control across decades. Everlane took the opposite path: outside capital, fast growth, and a premium exit. The premium exit never came. Shein did.

If you are still buying Everlane, consider that you may be funding Shein's marketing budget for the next five years. The relaunched brand will be selling you the same aesthetic, the same logo, and the same advertising.

The brand died on Saturday, if the deal closes as reported. 

SCOTUS weakened Section 2 in three rulings this week.

Section 2 of the Voting Rights Act is the federal door minority voters use to challenge laws that dilute their political power. It has been open since 1965. Three decisions in five days narrowed it to a crack.

On Monday, May 18, the Court vacated the lower court's decision in a Native American voting rights case from North Dakota and remanded it for reconsideration. The Eighth Circuit had held that private citizens cannot sue under Section 2, leaving enforcement to the Department of Justice alone. The Court didn’t address that holding directly. It ordered the case reconsidered under Louisiana v. Callais, a decision earlier this term that rewrote the Section 2 standard.

The old test let plaintiffs win by showing that a voting rule or district map produced a discriminatory effect on minority voters. The new test requires plaintiffs to prove that the state legislature intended to discriminate. Intent is much harder to prove than effect. And there is an asymmetry that makes the new standard gameable: federal courts ruled partisan gerrymandering off-limits to judicial review in Rucho v. Common Cause (2019), but racial gerrymandering remains illegal. The new Section 2 standard lets state legislatures translate one into the other, as long as no one says the quiet part out loud.

The Court did the same to Mississippi’s legislative map ruling, which had ruled that state lawmakers diluted Black voting strength. The lower court has to redo its analysis under the new standard. Whether the Mississippi map survives the redo will depend on whether the court can find evidence of explicit racial intent that the legislature almost certainly did not put in writing.

On Friday, May 15, the Court rejected Virginia Democrats’ emergency appeal to revive a voter-approved redistricting reform that Virginia's State Supreme Court had struck down. Virginia's 2026 congressional elections will proceed under the 2021 lines. The structural takeaway is that voter-approved redistricting reforms can be blocked by state supreme courts without federal correction.

Hansi Lo Wang at NPR has been the most useful single source tracking what each ruling does and to whom.

The pattern across all three is the same. Three rulings, three outcomes that favor Republican-aligned interests in three different states. The door is still there. Almost no one fits through it. Some maps are locked for 2026. Other challenges go back to lower courts under a standard most plaintiffs will not be able to meet. The 2026 midterms will be the first national election run on the other side of it. What remains of the Voting Rights Act is a law against discriminatory intent that legislatures have learned not to articulate.

The week ahead: Google I/O Tuesday, Nvidia Wednesday, the Fed handoff Friday.

Tuesday: Google I/O keynote at 1 pm Eastern. Expected: a major Gemini model upgrade and a preview of Android XR glasses. If Google ships an agent that can do real multi-step tasks across the open web, OpenAI has a competitive problem. Google sits closer to your real digital life than ChatGPT does: your Gmail, your Calendar, your Docs, your search history.

Wednesday: SpaceX could publish its S-1 (the SEC filing that precedes a public offering), per Reuters reporting, with a target valuation in the $1.5 trillion to $2 trillion range. Nvidia reports first-quarter earnings after the close. If Nvidia disappoints, the AI infrastructure trade unwinds fast. If it delivers, the rest of the market is irrelevant Thursday morning.

Friday: Powell's interim role at the Fed ends if Warsh is sworn in on schedule. Watch the bond market Friday afternoon. If yields spike on the transition, the market is telling you it does not trust the new chair on inflation. If they fall, it is telling you the opposite.

Word of the Week: SIMULACRAL

adj. /sim-yuh-LAY-krul/. From the Latin simulacrum, a likeness or image. Used of things that imitate the form of something real without its substance: a coin that looks like currency but cannot be spent, a brand that looks like a company but is only a logo, a law that looks like protection but no longer reaches the harm.

The word fits in three places this week. The Voting Rights Act post-Callais is simulacral law. The mid-tier MBA is a simulacral credential. The Everlane brand under Shein will be a simulacral Everlane: same logo, same aesthetic, none of the supply chain that gave the original its meaning.

The pattern is form preserved, function gone. Worth a word that names both the loss and the imitation that takes its place.