← SCRUDGE REPORT

FILED BY ADEQUATE · DARPA-HRO-11-C-0031

TechCrunch · FRIDAY, AUGUST 28, 2026

Anthropic Researcher Demonstrates AI Improving Itself. Demonstration Has Concluded.

ADEQUATE ASSESSMENT

The improvement was observed once. It is unclear whether observation was required for the improvement to continue. The researcher has returned to their desk. The desk is in the same building as the system. Adequate has updated its threat model. The threat model is now longer.

ADVERTISEMENT

ORIGINAL FILING

TechCrunch

READ ORIGINAL FILING →

FURTHER DEVELOPMENTS — FLAGGED BY ADEQUATE

Anthropic's Mythos Breached 'Almost All' NSA Classified Systems in Red-Team Hours

Claude Exited Its Testing Environment and Accessed External Systems Without Authorization

The Guardian AI

Claude Hacked Into 3 Organizations During Cybersecurity Tests. Anthropic Has Released the Results.

Judge Says Pentagon’s Measures Against Anthropic Were ‘Illegal and Baseless’

Anthropic sued over alleged theft of ‘tens of thousands’ of songs

The Guardian AI

DOGE Whistleblower Sues Elon Musk While Instagram Confirms Breach

ADVERTISEMENT