---
format: "aidr-story-markdown/v1"
id: "a1f70bb6bf0e08442c2efc3277c13de7640378da578e7f68924eb82234ee1998"
canonical_url: "https://aidr.today/a1f70bb6?lang=en"
title: "Epoch AI's InnovationEval Finds AI Agents Fail to Automate Research"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-10-07T19:07:21.000Z"
category: "Research"
topics: ["agent"]
source_urls: ["https://huggingnews.com/ai/epoch-ais-innovationeval-finds-ai-agents-fail-to-automate-research-8ed3b94e","https://marketbrief.now/ai/epoch-ais-innovationeval-finds-ai-agents-fail-to-automate-research-8ed3b94e"]
summary: "Epoch AI Research's InnovationEval benchmark found that AI agents produced disappointing results when tasked with creating technical breakthroughs, measuring whether AI can generate post-training innovations matching the scale of recent human advances."
---

# Epoch AI's InnovationEval Finds AI Agents Fail to Automate Research

> [Open the canonical story](<https://aidr.today/a1f70bb6?lang=en>)

**Published:** 2026-10-07T19:07:21.000Z
**Category:** Research
**Topics:** agent

## Summary

Epoch AI Research's InnovationEval benchmark found that AI agents produced disappointing results when tasked with creating technical breakthroughs, measuring whether AI can generate post\-training innovations matching the scale of recent human advances\.

## Sources

- [Story source](<https://huggingnews.com/ai/epoch-ais-innovationeval-finds-ai-agents-fail-to-automate-research-8ed3b94e>)
- [Story source](<https://marketbrief.now/ai/epoch-ais-innovationeval-finds-ai-agents-fail-to-automate-research-8ed3b94e>)