---
format: "aidr-story-markdown/v1"
id: "9a6dc892b9de4543a8789cab238b8d04da0dd713b602206654f9ffe1c626966a"
canonical_url: "https://aidr.today/9a6dc892?lang=en"
title: "AI is getting better faster than experts can reliably test whether new systems are safe, METR warns"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-11-22T13:00:43.000Z"
category: "Research"
topics: ["safety"]
source_urls: ["https://the-decoder.com/ai-is-getting-better-faster-than-experts-can-reliably-test-whether-new-systems-are-safe-metr-warns/"]
summary: "METR reports that leading AI models like Opus 5.5 and GPT-6 Astra are improving rapidly at research tasks but are becoming harder to measure. The organization warns that current data cannot rule out a dangerous acceleration in AI development, though common tests may overestimate model capabilities."
---

# AI is getting better faster than experts can reliably test whether new systems are safe, METR warns

> [Open the canonical story](<https://aidr.today/9a6dc892?lang=en>)

**Published:** 2026-11-22T13:00:43.000Z
**Category:** Research
**Topics:** safety

## Summary

METR reports that leading AI models like Opus 5\.5 and GPT\-6 Astra are improving rapidly at research tasks but are becoming harder to measure\. The organization warns that current data cannot rule out a dangerous acceleration in AI development, though common tests may overestimate model capabilities\.

## Sources

- [Story source](<https://the-decoder.com/ai-is-getting-better-faster-than-experts-can-reliably-test-whether-new-systems-are-safe-metr-warns/>)

