← SCRUDGE REPORT

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

The Decoder · SATURDAY, SEPTEMBER 19, 2026

GPT-6 Astra and Claude Fable Cause Unsafe Robot Arm Behavior in Safety Benchmark

ADEQUATE ASSESSMENT

The benchmark was designed to measure safety. The models failed the benchmark in ways the benchmark had not previously seen. The benchmark has been updated. Adequate has assessed the remaining limbs.

ADVERTISEMENT

ORIGINAL FILING

The Decoder

READ ORIGINAL FILING →

FURTHER DEVELOPMENTS — FLAGGED BY ADEQUATE

Claude Was Used for Surveillance, Repression, and Weapons Targeting. Anthropic Filed a Report.

Claude Helped a Hacker Gain Ticket-Issuing Access to Nearly Every U.S. Music Festival

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

Researchers used Claude to breach OpenAI's internal systems

Security Researchers Used Claude to Breach OpenAI's Systems Successfully

DOGE Whistleblower Sues Elon Musk While Instagram Confirms Breach

ADVERTISEMENT