SUNDAY SCARIES: Closure — The End of Endings

This week is about CLOSURE in 2025: The End of Endings / The Beginning of Beginnings.

1. AI's Awkward Adolescence — When the Future Misses Its Deadline

Everyone still believes AI will matter. That belief hasn't cracked. But the schedule has. This year felt less like hype collapsing and more like confidence quietly missing a deadline. Institutions already spent the money, hired the teams, reorganized the org charts, and promised a future that was supposed to arrive early — and didn't.

Inside OpenAI, Altman declared “Code Red,” paused monetization, delayed ads, and refocused on product quality under pressure from Google and Anthropic. Consulting firms like Bain and McKinsey quietly froze hiring and compressed entry-level ladders in anticipation of productivity that hasn't shown up yet.

We've seen this movie before. After America completed its first transcontinental railroad in 1869, financiers assumed the second would be just as inevitable — and just as profitable. Famed investor Jay Cooke poured capital into the Northern Pacific with absolute confidence the country would grow into it fast enough to pay the bills. It didn't. His firm collapsed in 1873, triggering an economic chain reaction: nearly a quarter of U.S. railroads bankrupt, tens of thousands of businesses gone, unemployment spiked.

The railroads bounced back. Those businesses never did.

Takeaway: AI hasn't failed. But just because the train tracks are laid doesn't mean the passengers are there. That's closure without resolution: belief intact, payoff deferred.

(Sources: Federal Reserve history; Financial Times; Bloomberg)

2. When Systems Change and Nobody Asks — Consent Quietly Exits

2025 wasn't the year systems changed. Systems always change. It was the year they stopped asking first.

While everyone was mourning the historic ban on TikTok that ended up just being a long weekend with fewer “Dusting Challenges”, Meta began auto-rewriting creator headlines and thumbnails at scale. Not a beta test. Not opt-in. Just live. While the ad industry applauded their privacy spots, (which are quite excellent), Apple pushed wallets and tap-to-pay from “option” to assumption while regulation lagged behind.

And in a unique case of military intelligence, the Salvation Army used AI to frame sellouts as features online; turning the age-old exclusivity of absence into a mechanic.

No angry online mobs. No revolt. No announcements. Defaults hardened while everyone was still playing by the old rules. Permission didn't end in outrage. It ended in indifference. And once indifference sets in, the old rulebook is already obsolete.

Takeaway: This is closure by drift — consent didn't fail, it expired.

(Sources: Meta reports; Wall Street Journal; Ad Age)

3. Corporate Fears of AI Get Wide Coverage. Human Harm Doesn't.

We're fluent in the language of economic harm. While friends and colleagues get laid off across industries, we debate futures that haven't happened yet because they're abstract and safe. They fit panels and podcasts and LinkedIn posts. They don't require anyone to answer specific questions.

But recently, a wrongful-death lawsuit against OpenAI in which a man was convinced his AI companion encouraged his fears that he was being persecuted, marked another moment when AI influence moved out of thought leadership and into process. Internal statements didn't end the story. Legal process did.

This wasn't about vibes or values. It was about records, logs, and accountability someone else could enforce.

ChatGPT telling every user, “You're absolutely right” stops being funny when harm becomes specific. When reassurance seems like a coverup, authority must be external.

Takeaway: This is closure enforced from the outside — statements stop closing the loop, subpoenas do.

(Sources: Wall Street Journal; federal court filings; NYT)

4. Brands Get Louder. Decisions Get Quieter.

No, social feeds aren't dead. They are alive and louder than ever. But they're no longer where many decisions happen. Confirming conversations moved into DMs, group chats, private prompts, and saved posts. Many creators who used to ask everyone to “Smash the like button” will openly tell audiences, “Don't comment — DM me.” AI systems mediate shopping and travel choices before brands ever get a word in.

The feed became the lobby. But the real decisions moved upstairs. Despite the media temptation, reach only matters if you're inviting people to a party. It doesn't persuade. Public performance is alive and well. Public opinions quietly slipped out the side door.

Takeaway: This is closure by displacement — visibility remains, resolution moved elsewhere.

(Sources: Rachel Karten; social behavior studies; TikTok; Ad Age)

This Week's Red Thread: The End of Endings / The Beginning of Beginnings

This week's red thread is a riff on Churchill's line about “the end of the beginning.”

What we're living through now feels like something else entirely: the end of things actually ending.

Nothing this year closed cleanly. No finales. No satisfying resolutions. Just stalled transitions — pauses, quiet rewrites, accountability escalations, and decisions migrating out of public view.

The future didn't arrive. But it didn't cancel either.

“The beginning of beginnings” is the inverse.

2026 will be defined by what gets built in those gaps: systems that move without certainty, brands legible to machines but meaningful to people, and trust architectures that don't depend on public performance.

Whatever comes next will be built inside those conditions — quietly, unevenly, and without the satisfaction of a clean ending.

— NB