The Decoder 🧠 Models 🕐 19h ago 👁 0 ⏱ 1 min read

Nine AI agents tried autonomous research on open questions from US national labs and none passed the experts' bar

The "Into the Unknown" benchmark tasks nine AI agents with tackling open research questions from U.S. national laboratories. None of the 135 submissions meets all requirements. A total of 114 papers fail the integrity check, with agents claiming to have conducted experiments that never took place. GPT-6 Astra performs best but still doesn't meet the experts' standards. The article Nine AI agents tried autonomous research on open questions from US national labs and none passed the experts' bar appeared first on The Decoder.

Source: The Decoder — Published November 7, 2026 at 11:13 UTC — Category: Models

🔗 Read full article on The Decoder →

More from The Decoder today

Anthropic pushes ahead with pre-Thanksgiving IPO despite tech selloff, accepts lower valuation 12h ago AI auditors catch most agent misbehavior but flood reviewers with false alarms, new study finds 13h ago Thinking Machines' Inkling 2 Small drops 316 F-bombs across benchmark runs 13h ago

Nine AI agents tried autonomous research on open questions from US national labs and none passed the experts' bar

More from The Decoder today

More in Models

Anthropic pushes ahead with pre-Thanksgiving IPO despite tech selloff, accepts lower valuation

AI auditors catch most agent misbehavior but flood reviewers with false alarms, new study finds

Thinking Machines' Inkling 2 Small drops 316 F-bombs across benchmark runs

The gap between open and closed AI models in cyber capabilities keeps shrinking

State propaganda in some Chinese AI models is only "a thin layer" that fine-tuning can strip away

Google's AI detects high blood pressure from a regular smartwatch without a cuff