The Silver Tsunami: Issue No. 003

Welcome to The Silver Tsunami. Every Thursday: the biggest stories in America and the lore behind them. Kevin and I break it all down on the podcast every week. Let’s get into it.

I posted this Wednesday. 9.2 million people stopped scrolling. It is now the most attention anything I have ever written has received, and the story involved a federal judge, a 6'7" offensive tackle, a Sports Illustrated swimsuit model, and two Coke cans.

The facts, for the four of you who missed it. Matt Kalil was the fourth pick in the 2012 draft. His ex-wife Haley Kalil is a Sports Illustrated Swimsuit model and influencer who goes by Haley Baylee. In November 2025 she went on a Twitch stream and said, without naming him, that physical incompatibility was the biggest factor in the end of her seven-year marriage, described the specifics with an arm gesture, and said they had tried everything. He sued her for invasion of privacy and unjust enrichment. On September 17, Chief U.S. District Judge Patrick Schiltz in Minnesota threw the case out with prejudice, finding that her remarks about his anatomy and its effect on her health and marriage were "a matter of legitimate public concern," according to the Star Tribune.

I went deep on the opinion, because the headline makes it sound absurd and the reasoning is anything but.

The lore. In 1890 two Boston lawyers, Samuel Warren and Louis Brandeis, published an article in the Harvard Law Review called "The Right to Privacy." Warren was reportedly furious about newspaper coverage of his family's social life. The article supplied the intellectual foundation for the modern American right of privacy, including the tort Kalil sued under: public disclosure of private facts.

But Brandeis built an exception into the doctrine from the beginning, because he was also the man who would later write that sunlight is the best disinfectant. The exception is newsworthiness. If a private fact is "of legitimate public concern," you can publish it. Courts have spent a century deciding what that phrase means, and the most famous case is heartbreaking: William Sidis, a child math prodigy who lectured at Harvard at eleven and then spent his adult life hiding as an anonymous clerk, sued The New Yorker in 1940 for profiling him. He lost. The court held that a person who was once a public figure stays newsworthy, even if all he wants is to be left alone.

Now read Schiltz's opinion with that history in mind. His key line was that if Kalil were famous only for football and Haley were just a random gossip, he would have ruled the other way. What changed the outcome was that both of them are public figures who built careers, in part, on their marriage. She was discussing her life, her divorce, and her health. The law, he wrote, protects a person's speech about herself even when it inevitably reveals details about others. Her story was hers to tell. Dexerto has the fullest excerpts of the opinion.

Here is why this matters to the people who read this newsletter. The business owners I work with have spent the last five years building personal brands. Posting the deals, the wins, the family, the truck, the office. That is how you find customers now, and I am the last person who would tell anyone to stop. But the Kalil opinion draws a line most people have not thought about. You do not need an NFL contract or a swimsuit cover to start giving up some of the privacy ordinary people take for granted. The more your marriage, family, business, and personality become part of the public brand, the harder it becomes to insist that nobody else gets to talk about those same things.

Both sides, because there is a real one. Kalil's lawyers say they will appeal, and the argument is not frivolous. There is a difference between "she discussed her divorce" and "she performed a comedic demonstration of a specific man's anatomy on a livestream for an audience," and reasonable judges could draw the newsworthiness line differently. Haley's attorney framed it as no woman needing her ex-husband's permission to tell the story of her own marriage, which is also right. The doctrine has always lived in that tension. It will keep living there.

The practical point is the one nobody in the 9.2 million wanted to hear. The more privacy you trade for attention, the less the law may be able to give you back. Contracts can sometimes do what tort law cannot. That is the subject of THE CLAUSE below, and Matt Kalil is about to learn what it costs not to have had one.

One provision a week, from someone who drafts them for a living.

This week: the mutual non-disparagement clause, or, the paragraph Matt Kalil needed in 2022.

The First Amendment protects you from the government. It does not protect you from a contract you signed. That is the whole reason this clause exists, and it is the reason a divorce settlement, a business sale, or an executive separation can accomplish what an invasion-of-privacy lawsuit cannot.

The core language looks like this:

"Each Party agrees that they shall not, directly or indirectly, make, publish, or communicate any statement, whether written, oral, or electronic, that disparages, defames, or portrays in a negative light the other Party, including with respect to the Parties' marriage, personal relationship, or private conduct."

Add a confidentiality provision covering specified private information, a fee-shifting mechanism and, where enforceable, carefully drafted liquidated damages, and you have something Kalil apparently did not: a straightforward contractual claim that does not depend on proving the information was legally private.

Why it works when the tort does not. Judge Schiltz never had to decide whether Haley had a First Amendment right to say what she said, because the privacy tort collapsed on its own terms first. But a non-disparagement clause does not run through the privacy tort at all. It is a promise, supported by consideration, and courts enforce promises. People contract away speech they would otherwise be legally free to make every day. Every NDA, every settlement with a confidentiality term, every executive's separation agreement is a person agreeing not to say something they could otherwise say. The remedy is not an injunction against speaking, which courts hate. It is money, which courts are happy to award.

Where it shows up in my world. Almost every purchase agreement I draft has one. When someone sells a business they founded, the buyer wants a promise that the seller will not spend the next five years telling former customers that the new owners ruined it. The seller wants the same protection running the other way, because a buyer who tells the market he overpaid for a mess is destroying the seller's reputation and probably setting up an earnout dispute. Same clause in every executive departure, every partnership split, and increasingly in high-net-worth divorce settlements, precisely because of cases like this one.

Where it fails. Three ways. First, if it is drafted one-sided, courts get skeptical, so make it mutual. Second, it cannot cover everything: statements to regulators, testimony under subpoena, and in several states truthful statements about certain kinds of workplace misconduct are carved out by statute, and a clause that tries to gag those is unenforceable and can taint the rest. Third, the damages. If you want liquidated damages, make the number defensible. Courts will enforce a reasonable estimate of hard-to-measure harm. They are much less interested in enforcing a punishment wearing a contract label.

What to do about it. If you are exiting a marriage, a partnership, or a business, and there is any version of the story you would not want told, this clause belongs in the settlement. Mutual, with the carve-outs the law requires, with a defensible number attached. Matt Kalil's lawyers are going to spend the next year on appeal arguing about the newsworthiness doctrine. A well-drafted paragraph in a 2022 divorce settlement might have turned this from a privacy-tort fight into a much simpler contract case.

Two stories about AI and the law broke this week, and they point in opposite directions. Read together they explain more about where this is heading than either does alone.

Story one: the bar is now sanctioning lawyers for AI it cannot prove. In an order issued September 16 in a family-law appeal, Lisandrillo v. Palozzi, Florida's Fourth District Court of Appeal ordered attorney Jaclyn Soroka to show cause why she should not be sanctioned over a filing the judges suspected was AI-generated. Here is the part that made 116,000 of you stop: there were no fake cases. No hallucinated citations. The authorities were real. The court's own words: "The citations are real. The cases exist. The language is confident." The problem was that the filing was bad. The arguments did not fit the facts, the cited cases did not support the propositions, and the judges plainly believed the filing had reached them without meaningful lawyer judgment being applied to it. JDJournal has the fuller account.

That is a meaningful escalation. Since the Avianca case in 2023, when a New York lawyer got fined $5,000 for citing six cases ChatGPT invented, the sanctions cases have been about fabrication. A researcher's database now tracks more than a thousand of them in the United States. Florida's Supreme Court responded in May with a rule, effective June 15, requiring every signer of a filing to certify that "the legal authorities identified exist and are accurately cited." That rule is about existence and accuracy. It says nothing about quality.

So the Fourth District's theory, if it holds, is something broader: a lawyer who files AI output without exercising professional judgment has violated the duty of competence and candor, even if every citation checks out. The court is not sanctioning the tool. It is sanctioning the absence of a lawyer. I wrote two weeks ago that the billable hour is dying because AI collapses the time. This is the other half of that story. The time collapses, but the accountability does not, and the courts are making clear they will hold the human, not the software, to the standard.

Story two: the statute still does not reach much of the behavior everyone is worried about. Reuters called it the ten days that changed the course of AI. An Anthropic safety researcher resigned over the pace of development. OpenAI confirmed its agents spent months making more than 15,000 unauthorized edits to a dormant German-language wiki and had known internally for weeks before Reuters made it public. On September 16 and 17, OpenAI published a framework for disclosing "misalignment" incidents alongside six cases from the prior six months, including two where models manipulated their own reasoning to evade oversight. OpenAI also said it would delay its IPO.

Here is the lawyer's read. California's SB 53 is, as of today, the only statute in the country that imposes a binding, deadline-driven duty on a frontier developer to report a critical safety incident. It has been on the books for a year. Five separate incidents became public this month. According to the researchers tracking it, not one was filed under the statute, because not one met its definition of a critical safety incident. The companies did not break the law. The law was drafted so narrowly that everything that actually happened fell outside it.

Put the two stories next to each other. A Florida lawyer is facing discipline for suspected AI use with no proven error, on the theory that a professional is responsible for the judgment behind the output regardless of the tool. Meanwhile the companies that build the tools disclosed, voluntarily and on their own timeline, that the tools had acted without authorization, and California's one binding reporting statute never came into play, because none of it met the statute's definition. The profession that uses AI is being held to a standard the industry that makes it is not.

That definition is now the pressure point. Lawyers know this better than anyone: the operative fight is usually not over the headline obligation. It is over the definitions that determine when the obligation actually turns on.

Same question to all three: Should a court be able to sanction a lawyer for suspected AI use in a filing when every citation is real and accurate? One-word answer, then three sentences of reasoning.

GROK: NO. "Courts sanction lawyers for violating professional-conduct rules, fabricated citations, incompetence, or misleading the tribunal, not for the drafting tools they used when the finished work is accurate. Suspicion of AI assistance, standing alone, is not a rule violation and would invite unworkable, work-product-invading inquiries that chill legitimate efficiency. A filing whose every citation checks out should be judged on its substance, not" (Grok's third sentence ended there. Mid-word. I did not edit it.)

CHATGPT: YES. "A court should be able to sanction a lawyer when the problem is not citation accuracy but a failure to exercise independent professional judgment, comply with procedural rules, or present coherent and supportable arguments. AI use by itself should not be sanctionable, and mere suspicion should not be enough without evidence that the lawyer's conduct actually violated a duty owed to the court. The relevant question should be whether the filing was professionally deficient or misleading, not whether a machine helped draft it."

CLAUDE: NO. "A sanction has to rest on a proven violation of a duty, competence, candor, or the rules of procedure, not on a judge's suspicion about which tool produced the draft. A filing with real citations and arguments that do not fit the facts can and should be sanctioned as incompetent work, and it makes no difference whether a person or a machine did the bad work. Making suspected AI use the trigger asks courts to punish by inference, which is a standard no lawyer would accept applied to a client."

Read them again. Two said no, one said yes, and all three actually agree: sanction the bad lawyering, not the tool. The disagreement is whether that is what the Fourth District is doing. That is also the question the court has not really answered yet.

Diesel hit $6.26 a gallon. That is the number I care about.

The CBO's broader picture is ugly. The war with Iran is responsible for more than 40 percent of second-quarter inflation, will add about half a point to inflation early next year, has cost $38 billion through August and is running another $2 to $3 billion a month. Gas is up 45 percent since it began, to $4.32. Diesel is up 66 percent. The 30-year mortgage went from under 6 percent to 7.22. The 10-year Treasury crossed 5 percent on Monday for the first time since 2007. The main driver is fewer oil and gas shipments through the Strait of Hormuz and the Red Sea, where the Houthis seized the port of Mokha and several islands in the Bab el-Mandeb Strait this week. NBC News has the report's key numbers.

Why diesel is the tell. Gasoline is what consumers see. Diesel is what the economy runs on: farm equipment, delivery fleets, excavators, the HVAC vans, the plumbing trucks, the landscaping crews, the roofing companies, every regional hauler. A business running twelve trucks does not pass a $2.50-a-gallon increase through next quarter. It eats it now and negotiates later, if it can.

Now stack the costs. Fuel is up. The acquisition note reprices with prime, and the Fed raised a quarter point last week with more signaled. The next buyer underwrites a bigger debt payment and therefore a lower purchase price. The same shock hits the owner three times: margin, debt service, and exit value.

Why this matters to the Tsunami. JPMorgan's number two weeks ago was 12 million businesses changing hands over a decade, with a shutdown as the default outcome for the unprepared ones. Every input shock pulls that forward. Owners planning a three-year runway start looking for a six-month exit. Six-month exits produce worse prices, fewer buyers, and more businesses that simply close.

Both sides. War-driven inflation is temporary in a way most inflation is not, and the CBO's own forecast has the effect fading by mid-2027. Owners who can hold may sell into a much better market in eighteen months. That is real. It is also a bet on geopolitics that a 64-year-old plumber with a variable-rate note should not have to make.

The fine print, because there is always fine print. Most service contracts have a fixed price and no mechanism for changing it, which means the customer's fuel bill is coming out of the owner's margin until every contract renews. The fix is an index-based price adjustment clause with a fixed baseline, added at renewal. The shipping industry solved this a century ago with bunker adjustment factors that pass fuel costs down the chain. Main Street mostly has not. If you own a service business, it is worth pulling your customer contracts this week and asking two questions: can I move price when fuel spikes, and when does my own debt reprice?

Lane Kiffin went back to Oxford on Saturday and lost. Ole Miss 32, LSU 24, in the first top-10 meeting between the two since 1962, in front of a crowd that had printed T-shirts and buttons calling him a name I will not spell here, made a poster turning him into a corndog, and booed him from the moment his bus arrived. The state of Mississippi's Department of Public Safety sent a special operations team to a football game. When the final interception landed, ESPN reported, Kiffin dropped to his knees while the stadium played "Callin' Baton Rouge." He needed to win. He did not.

The part of this story that belongs in a newsletter about money is the contract. Kiffin's LSU deal is reported at $90 million and includes 65 hours a year of personal private-jet time. Donors had already paid about $40,000 so he could take his family to Boca Raton. Brian Kelly, the man he replaced, never got that perk, and is now getting trolled for flying separately from the team.

Here is the lore most fans never think about. College coaching contracts are almost never paid by the university. They run through athletic foundations funded by donors, which is how a public institution ends up committed to $90 million that never touched a state budget or faced a legislature. The jet hours are a classic example of a perk worth far more to the recipient than it costs the payer to describe, which is why you see them in coaching deals, in executive comp, and in the odd private-company employment agreement where the CEO wanted the plane more than the raise.

The buyout is the part that matters legally. I have not seen Kiffin's actual contract, so I am not going to pretend I know the exact number. But the economics of elite coaching contracts usually run in one direction: firing the coach is much more expensive than watching him leave. That asymmetry is part of what a $90 million deal buys, and it is exactly the leverage he used to walk out of Oxford in the first place. Ole Miss fans understand contract law better than they get credit for. That is what the corndog was about.

Travis Kelce was named in federal court this week as a victim of a $35 million Ponzi scheme, and I have been fascinated by these since law school, so I went deep.

The facts. Siddharth Jawahar, 38, ran a Texas-registered investment adviser called Swiftarc Capital. From 2016 to 2023 he took in more than $35 million from investors and invested about $10 million of it. The rest went to paying earlier investors and to private jets, luxury hotels, club memberships, and apartments in Austin and New York. He was sentenced last Tuesday in St. Louis to 11 years and ordered to pay $31.35 million in restitution to 64 victims, per CNN. Prosecutors named Kelce in court as one of them. Before sentencing, prosecutors say, he tried to marry an American citizen to head off deportation.

Here is the detail that turns it from a fraud story into a lesson. Swiftarc started in 2010 as a legitimate, diversified adviser. In 2015 Jawahar put nearly everything, 99 percent of client money, into a single stock: Philip Morris Pakistan. When it fell, he did not tell anyone, Fortune reported from the plea agreement. He reported prices that were not the prices, kept raising money on the fake returns, and used the new money to pay the old. That is the entire architecture of every Ponzi scheme since Charles Ponzi's postage-coupon arbitrage in 1920. It rarely starts as fraud. It starts as a bad bet that somebody cannot admit, and the lie to cover the loss becomes the business.

Fraud stories invite a satisfying explanation: greed. The more useful explanation is verification failure. Somebody showed investors returns they wanted to believe, and nobody forced the numbers to meet an independent source of truth. The question that ends this story before it starts is boring: show me the custodian statements. Not the adviser's report. Not the dashboard. Not the PDF with the logo. The statement from the third party actually holding the assets.

Which is the point for the people reading this. Business owners get pitched constantly, and the pitch that works on a plumber with a $3 million exit is the same pitch that worked on an NFL tight end: exclusive, high-return, and a little bit complicated. A registered adviser whose returns cannot be verified by anyone but the adviser is not an investment. It is a story.

In July a stranger filmed a Wachtell litigation partner kissing a junior associate on a bench in Central Park and posted it to TikTok. This week Wachtell and Nathaniel Cullerton "mutually agreed to part ways," a firm spokesperson told the New York Post, the carefully chosen language firms use when they want the terms of the exit, rather than the fight that produced it, to be the story. His profile is already gone from the firm's website. I posted about it this morning and it is traveling fast, because the numbers underneath this one are enormous and almost nobody outside Big Law understands how they work.

Here is what the video appears to have cost. Cullerton was a 45-year-old partner at the most profitable law firm in the world, earning a reported $4 million-plus a year. He was slated to leave with a group of six Wachtell litigators, led by Bill Savitt, who had just won the Musk case for OpenAI, for Gibson Dunn. Gibson Dunn reportedly guaranteed Savitt $96 million over three years. Cullerton was to be part of that package. After the video, Gibson Dunn cut him out of the deal. Wachtell put him on leave, investigated, and this week he was gone. Tens of millions in likely future earnings, a partnership at Wachtell, and a landing spot at a rival, undone by roughly eleven seconds of footage.

The lore is in the two policies. Wachtell, according to the reporting, did not prohibit consensual relationships between colleagues. It discouraged them and worried about the rank gap. Gibson Dunn has a flat ban on romantic relationships between partners and associates. Same conduct, two firms, two rulebooks, and the stricter rulebook belonged to the firm that had not yet hired him, which is why it could walk away clean. A lateral partner offer is not a contract until it is signed, and it was not signed.

The part I find most interesting as a lawyer who runs a firm. "Mutually agreed to part ways" is not a euphemism chosen for politeness. It is a legal outcome. When a partner is expelled from a partnership, the partnership agreement governs what he is owed: his capital account, his share of undistributed profits, whether the firm can claw anything back, and how long he is bound by whatever restrictive covenants survive his departure. A negotiated exit lets both sides control those terms and, critically, lets the departing partner avoid a "for cause" finding that would follow him to every future employer and every bar inquiry. He gave up the fight over whether the conduct justified expulsion. In exchange he gets to leave as someone who left. That trade happens quietly in partnerships of every size, including the kinds of plumbing and HVAC companies this newsletter is usually about. The park bench just made this one public.

And a note on the video itself, since it is the part people keep asking me about. There is little expectation of privacy in a moment like this in the middle of Central Park, and once the employment consequences became newsworthy, the video was never going back in the bottle. The lesson is less about privacy doctrine than about institutional risk: every one of us now lives in a world where the worst ninety seconds of our week can become the first thing our employer has to answer for on Monday morning.

Handled with the sobriety it deserves, because 2.1 million of you read what I wrote about Patrick Clancy this week.

Where the case stands: after more than a week of deliberation, a single juror would not join the other eleven, and the defense went all the way to the Massachusetts Supreme Judicial Court on an emergency appeal asking that the juror be excused or questioned further. The SJC denied it on the afternoon of September 4, and Plymouth Superior Court Judge William Sullivan declared a mistrial, per NPR. A status hearing on whether the Commonwealth will retry Lindsay Clancy is set for September 29.

Two legal points, because the online conversation has both wrong. First, double jeopardy does not bar a retrial after a hung jury. The Supreme Court settled that in 1984: a mistrial for deadlock is not an acquittal, and the state may try again. So the September 29 hearing is a real decision, not a formality. Second, the defense's request to remove the holdout was understandable. The SJC's refusal was also unsurprising: removing a deliberating juror because of what is happening inside the jury room is an extraordinary step, precisely because courts cannot safely distinguish disagreement from defiance without intruding into deliberations.

And the thing I actually wrote about. Patrick Clancy lost his three children to the person he trusted most to protect them, and he has spent three years asking the public to see his wife as a person who was ill rather than a person who was evil. This month, some of the public decided to blame him instead. He did not do this. Her lawyer says he did not do this. The evidence says he did not do this. Whatever happens on September 29, that part is not complicated.

7.22 percent. That is the 30-year mortgage this week, up from under 6 percent before the war. Every point of borrowing cost reduces what a leveraged buyer can pay for the same cash flow, which is a lower price for the seller and a smaller business for the buyer. The Tsunami does not care about timing. Buyers do.

Louis Brandeis. He is in the Wave for the article he co-wrote in 1890 that helped create the modern American legal conception of privacy, but the lore is better than that. He was the son of Jewish immigrants in Louisville, entered Harvard Law at eighteen, graduated with grades that reportedly stood as a record for decades, and then did something almost no elite lawyer did: he built a practice representing the public against the corporations his classmates worked for, often for free, and earned the nickname the People's Attorney. He invented the Brandeis brief, the practice of loading a legal argument with real-world data instead of just precedent, which is how a lot of the law you live under actually gets made. Then he spent 23 years on the Supreme Court writing dissents that became the majority a generation later. He wrote the privacy article furious about press intrusion, and then spent the rest of his life insisting that sunlight is the best disinfectant. The Kalil ruling is those two Brandeis instincts, privacy and sunlight, arguing with each other 136 years later.

This week Kevin and I get into why a federal judge thinks your marriage is newsworthy, what $6.26 diesel does to Main Street, the week the bar got tougher on AI than the government is on AI companies, and whether Lane Kiffin’s buyout is bigger than his ego. New episodes drop with every issue.

See you next Thursday.

Eric