When J.P. Morgan announced its five-day return-to-office mandate in January 2025, employees pushed back hard. More than a thousand signed a petition arguing the policy "disproportionately" pushed out women, caregivers, senior employees, and people with disabilities. At a February town hall, Jamie Dimon gave the response that became immediate legend: "I don't care how many people sign that f—ing petition." He later apologized for the language, but never for the position. The mandate took effect in March. Goldman Sachs had staked out similar ground under earlier leadership; former CEO David Solomon called remote work "an aberration that we're going to correct as quickly as possible" and never publicly walked it back. Both firms built their employer brands on the idea that in-office attendance was non-negotiable.

And then, the World Cup came to New Jersey. According to internal memos reported by the Financial Times, both Goldman and JPMorgan are now temporarily allowing employees to request remote work on match days during the 2026 tournament. The reason isn't some new productivity research, which hasn't changed. It isn't the caregiving argument employees made in writing and were told didn't matter. Instead, it's that hundreds of thousands of football fans will be swarming New York and New Jersey, transit will be disrupted, and workers who commute on impacted routes may simply not be able to get in. Soccer has changed the trains, and the trains have changed the policy.

Amazon is not doing this. Per Bloomberg, Amazon is directing workers to come in earlier than usual to beat the transit chaos and flagging alternative route options. One reading of Amazon's position is principled consistency; if five days a week is actually about collaboration and performance, a transit disruption doesn't really change the underlying rationale. Another reading of course is that Amazon understands any exception, whatever the reason, becomes the opening frame for the next argument. Both readings are defensible but what's harder to defend is the sequence underneath the banks' version. Goldman and JPMorgan employees were NOT granted flexibility when they petitioned on caregiving. They weren't granted it when hybrid productivity research accumulated. They're getting it now because the trains aren't running. The message that sends is unambiguous: the rule was never really about whether the work can be done from home but about who gets to decide, and on what grounds.

Now, so far, this is costing them nothing. The coverage is all outside-in; Fortune, Inc. Magazine, the trade press noting the irony with a raised eyebrow ("Soccer, apparently, is a different matter"). But there's no second petition, no internal uproar being reported, no walkout. The same workforce that mobilized over the mandate has, as far as anyone can tell, basically shrugged at the contradiction. The banks declined to comment, and they can afford to, because a temporary, logistically-justified exception that happens to benefit employees doesn't generate the same heat as a permanent policy imposed against their objection.

So here's what to actually do with this when your own CEO is about to reverse a position, make an exception, or break from something you've said publicly. Before they speak, ask: who comes out ahead from this change, and who comes out behind? Then just sort it through honestly. If the people most affected are getting something they wanted, you can let the CEO be direct about it; name the real reason, even if it contradicts the old line, because the contradiction lands as a win and nobody picks a fight over a win. That's the World Cup situation, and the comms move is to stop overthinking it and just say the plain thing: the commute is going to be brutal, work from home those days.

But if the people most affected are losing something like a benefit, flexibility, a promise, and the new position takes it from them, that's the petition situation, and the job is the opposite. Slow the CEO down before they say anything dismissive. The damage from Dimon's "f—ing petition" wasn't the policy as much as it was the contempt layered on top of a real loss. Your job there is to make sure the person delivering the news names the cost out loud, explains the trade-off in terms the affected group would recognize as fair, and never waves away what people are losing. You won't make them happy but with any luck you will keep a defensible decision from curdling into a resentment story.

The practical test, in one line you can use in any prep meeting: don't ask "are we being consistent?" Ask "who has to pay for this, and do they know we know?" That tells you in thirty seconds whether you're handing out a gift or detonating a grievance and which one you're doing changes everything about how the CEO should open their mouth.

The standard playbook for data center opposition runs like this: identify the affected community, meet with local officials, publish an economic impact report, host a town hall, tout the jobs created, promise to minimize the environmental footprint. It's a stakeholder engagement approach built for a localized problem. But according to a new survey of 6,872 registered voters by Milltown Partners, published Monday, the data center opposition problem is no longer localized.

Only 8 percent of the people who say they oppose data centers say they even know of one near their home. The way I read that is that almost everyone objecting is objecting to something they've never personally encountered. And the reasons they give aren't about their neighborhood. The survey found 67 percent pointing to rising energy bills and 59 percent to the sense that AI's gains will land on corporate balance sheets rather than in their pockets. Those are national affordability and equity arguments, not local nuisance complaints. As one analyst tracking the backlash put it, what people oppose isn't the building, it's AI itself, and stopping the data center is how they think they stop AI. The community relations playbook doesn't yet have an answer for that. Meanwhile, in the first quarter of 2026 alone, public backlash delayed or blocked at least 75 data center projects worth a cumulative $130 billion, close to the total lost across all of 2025.

Meta appears to have read this as a local problem. Last year the company spent more than $6 million on a folksy ad campaign across eight states and Washington, D.C., highlighting the economic benefits data centers bring to local communities. That campaign does real work for the slice of the country that lives near a data center and was perhaps on the fence. The much larger group opposing data centers on principle (the rising electricity bill, the suspicion that AI's upside is already being captured somewhere above them) is not watching those ads and reconsidering.

So if your company is building AI infrastructure, deploying AI, or has any meaningful AI story to tell publicly, the question your CEO will eventually face is "what is AI doing for people who don't have power?" The local stakeholder engagement model isn't calibrated for an ideological argument and right now the ideological argument is the one blocking $130 billion worth of projects in a single quarter.

On Tuesday, the AI trade had its worst day in months. South Korea's KOSPI closed down roughly 10 percent, with memory-chip giants Samsung Electronics and SK hynix each falling more than 12 percent and triggering circuit breakers that briefly halted trading in Seoul. The contagion spread; U.S. memory names like Sandisk and Micron Technology dropped, and the Nasdaq finished the day down 2.2 percent. What's worth noticing for our purposes is how unsettled the explanation is. Reporters and analysts pointed in several directions at once - leadership departures at Google stoking AI-competitiveness fears, regulatory scrutiny of leveraged chip-linked financial products, nerves ahead of Micron's earnings, even a Bank of America note warning of possible rate hikes. CNN's markets desk noted there was no single obvious catalyst. The one narrative that is traveling fastest and farthest though, was the simplest which is that the AI investment thesis is being stress-tested in public.

For employees, the precise cause barely matters. The headline they see is "AI selloff," and they're reading it against everything they've been told internally for the past 18 months. Because here's the thing communicators sometimes forget: at a lot of companies, AI wasn't just the growth story for investors. It was also the explanation given to employees for the hardest decisions like why roles are changing, why colleagues were let go, why capital spending is climbing while headcount falls. AI was the reason. So when the market starts publicly questioning the thing you told people justified cutting their teams, that lands differently than a normal bad day on the tape. Investors recalibrate price targets. Employees recalibrate whether the sacrifice was worth it and unlike investors, they can't sell the position.

That's the bind of this moment, and it's worth getting ahead of before the next all-hands. Not with a fresh round of conviction about AI's long-term payoff because your people have heard that, and right now the screen is arguing the other side. The more useful question is whether you've built enough credibility that, when the story wobbles in public, your employees extend you the benefit of the doubt rather than concluding the layoffs were sold to them on a premise the market no longer believes. You can't manufacture that credibility in the week the stock drops. You either banked it earlier, or you didn't.