# AI-2027 Reality Tracker

> Independent tracker comparing predictions from the AI-2027 scenario (ai-2027.com) against documented real-world events. Updated 29 March 2026.

The AI-2027 scenario was published in April 2025 by Daniel Kokotajlo, Scott Alexander, Eli Lifland, Thomas Larsen, and Romeo Dean (AI Futures Project). It predicts the trajectory from current AI agents through superhuman coders (March 2027), intelligence explosion (mid 2027), and potential loss of human control (late 2027).

This site tracks which predictions have been confirmed, partially fulfilled, or diverged from reality.

## Pages

- [English version](https://ai-2027-timeline.online/en.html): Full tracker with interactive graph and sourced timeline
- [Spanish version](https://ai-2027-timeline.online/es.html): Spanish translation
- [Czech version](https://ai-2027-timeline.online/cs.html): Czech translation

## Status as of March 2026

Confirmed or partially fulfilled predictions:
- Stumbling AI agents by mid 2025 (fulfilled: Claude Code, Cursor, Copilot agent mode)
- Massive datacenter buildout (fulfilled: ~$700B committed by Big Tech for 2026)
- Coding automation / 50% AI R&D speedup (partial: 2x coding uplift reported, broader R&D impact unclear)
- China closing capability gap through extraction (confirmed: 16M+ distillation attacks, $2.5B GPU smuggling)
- Government contracting AI labs for military (confirmed: $200M Pentagon-Anthropic contract, Jul 2025)
- AI deployed in active military operations (confirmed: Claude used in Venezuela raid and Iran strikes)
- Autonomous military systems (confirmed: robot-on-robot combat in Ukraine, Mar 2026)
- Government asserting control over AI labs (confirmed: Anthropic banned from federal contracts, Feb 2026; court blocks ban as unconstitutional retaliation, Mar 2026)
- Labs prioritizing coding/R&D as the critical capability (confirmed: OpenAI shuts Sora, pivots to coding, Mar 2026)

Unresolved predictions:
- China nationalizes AI research (mid 2026)
- AI takes measurable share of white-collar jobs (late 2026)
- Superhuman coder achieved internally (March 2027)
- China steals model weights (February 2027)
- Intelligence explosion (mid 2027)
- Misaligned superintelligence (late 2027)

## Key finding

The technical timeline (superhuman coder by March 2027) remains unproven. But geopolitical, institutional, and military dynamics are tracking with uncomfortable precision, and in several cases reality is ahead of the scenario's schedule.

## Independence

This tracker is independent and not affiliated with the AI Futures Project or any of its authors. All interpretations are the author's own. All claims are sourced. The original scenario and all credit for the predictions belongs to the AI-2027 team at ai-2027.com.