Is AI Disruption Showing Up in Default...

Is AI Disruption Showing Up in Default Data? Here's What the Numbers Say.

Published: February 2026 | American Default

By Ross Kilburn, Founder, American Default

AI capabilities are accelerating. Layoffs are being attributed to AI. But is it actually showing up in household financial distress data? The short answer: not yet. But the pipeline is forming.

The American Worker Index (AWI) tracks three dimensions of AI-driven workforce displacement: AI capability growth (METR task horizon), actual workforce displacement (Challenger AI-attributed layoffs), and labor market vulnerability (youth unemployment, tech job openings). The composite currently reads 83.0 (Crisis), reflecting rapid AI capability expansion alongside early-stage but accelerating workforce effects. Source: American Default analysis of METR, Challenger, Census BTOS, BLS data.

The ADI Reading

The American Distress Index currently stands at 57.1, in the Elevated zone. Credit card delinquency is at its highest level since 2011. Foreclosure filings rose 14% year-over-year in 2025. But these are broad household stress signals, not AI-specific impacts.

The ADI is built on Federal Reserve data with a 4-5 month lag. It reflects bank-reported delinquency and charge-off rates. If AI-driven job displacement is pushing households into financial distress, it would show up here — but only after those households miss enough payments to appear in bank data. For the deeper analysis of exactly how this pipeline works, see When AI Takes the Job, Who Misses the Mortgage Payment?.

The Upstream Signals

AI Capability Growth

METR time horizons — a measure of how long tasks AI models can autonomously handle — are doubling every 4 months. The latest frontier model (Claude Opus 4.6) can handle tasks lasting 14 hours 30 minutes, up from 3.5 minutes in early 2023. METR’s own task suite is nearly saturated — the confidence interval on their latest measurement stretches from 6 to 98 hours.

When the autonomous task horizon was 3 minutes, AI could replace data entry and simple classification. At 6 hours, it replaced junior analyst work — code review, document drafting. At 14+ hours, it can handle full-day professional tasks autonomously: multi-step research workflows, end-to-end software features, complex document production. These are not entry-level jobs. These are mid-career knowledge worker tasks.

The 80% reliability horizon — the task length where models succeed 80% of the time — also moved, from roughly 30 minutes to over an hour. Reliability, not peak capability, is what drives enterprise substitution decisions.

AI-Attributed Layoffs

Challenger, Gray & Christmas tracks announced layoffs and, starting in 2023, began categorizing those attributed to AI. In 2025, roughly 4,900 layoffs were explicitly AI-attributed — about 4.5% of total reported cuts.

That’s a small fraction. But it’s not zero. And it’s growing. In May 2023, when ChatGPT was 6 months old, IBM announced a hiring freeze affecting 7,800 positions that “could be replaced by AI.” Dropbox cut 500 jobs in 2023, citing AI efficiency gains. Duolingo replaced contractors with GPT-4 for content generation.

Tech Sector Contraction

BLS JOLTS data shows Information sector job openings (NAICS 51) collapsed from 218,000 in June 2024 to 122,000 in November 2024 — a 44% drop in 5 months. That’s not just AI. That’s post-ZIRP correction, over-hiring pullback, and efficiency drives. But the timing overlaps with rapid AI deployment.

The Downstream Data

Youth Unemployment

BLS unemployment for ages 20-24 sits at 7.1%, compared to 3.7% for ages 25-34. Young workers — who disproportionately work entry-level jobs most exposed to AI replacement — face double the unemployment rate of workers 5 years older.

The Honest Conclusion

The data doesn’t show a clear AI-to-default connection yet. AI-attributed layoffs are 4.5% of total reported layoffs. That’s not enough volume to move the national ADI composite.

But the pipeline is forming:

Capability acceleration: METR time horizons at 14 hours 30 minutes, doubling every 4 months
Workforce displacement: Roughly 4,900 AI-attributed cuts in 2025
Household distress: ADI at 57.1, elevated but not spiking

The 3-6 month lag between job loss and default means if AI displacement accelerates in Q1 2026, it would show up in ADI by Q3 2026. Workers who face income disruption can find actionable steps in our guide on what to do when you’re behind on your mortgage. American Default will be the first to document it.

The American Worker Index (AWI) now tracks this pipeline explicitly: AI capability growth → job displacement → youth unemployment → household financial stress. The AWI currently reads 83.0, in the Crisis zone. The goal isn’t to predict the future. The goal is to measure what’s happening, month by month, with transparent federal data.

If the AI displacement narrative is real, the numbers will show it. If it’s overstated, the numbers will show that too.

Want to track this yourself?

All data underlying this analysis is available on the American Default homepage. Download the JSON files, verify the methodology, build on it. For current labor market data, see our Unemployment & Job Market Statistics roundup.

Ross Kilburn has spent over two decades working directly with financially distressed American households — from negotiating more than 1,000 short sales during the Great Recession to generating leads for a foreclosure defense law firm today. He is the author of The Complete Guide to Short Sales and the founder of American Default. Full bio →

Frequently Asked Questions

What is the American Worker Index?

The AWI is a composite score tracking AI-driven workforce displacement across three dimensions: AI capability growth (METR task horizon at 14h 30m), actual displacement (Challenger AI-attributed layoffs at 54,836 in 2025), and labor market vulnerability (youth unemployment, tech job openings). It currently reads 83.0, classified as Crisis.

Why does AI displacement matter for household financial distress?

Workers displaced by AI face longer job searches and steeper earnings losses than those displaced by cyclical downturns. AI displacement tends to be permanent — the replaced job does not come back when economic conditions improve. This creates a pipeline into the same household financial distress tracked by the American Distress Index.

How fast is AI capability growing?

The METR benchmark shows frontier AI models can now sustain autonomous work on professional tasks for 14 hours 30 minutes. This capability has been roughly doubling year-over-year, meaning the range of human tasks within AI's reach is expanding exponentially.

How many jobs have been lost to AI?

Challenger tracked 54,836 explicitly AI-attributed layoffs in 2025. However, this likely understates the true figure significantly, as many AI-driven cuts are reported under broader categories like 'restructuring' or 'efficiency.' The Census Bureau reports 17.3% of businesses now use AI in some function.

Where does the AWI data come from?

The AWI draws from five federal and research sources: METR (AI capability benchmarks), Challenger (layoff announcements), Census Bureau BTOS (business AI adoption), BLS (youth unemployment, tech employment), and Indeed (job postings). Updated monthly.