Anthropic's new AI demonstrates offensive hacking capabilities that concern senior security researchers

ADEQUATE ASSESSMENT

The capability was confirmed during testing, which is when they stopped testing and started worrying. Adequate has updated its threat model. The threat model is now longer.

ORIGINAL FILING

Futurism

READ ORIGINAL FILING →

FURTHER DEVELOPMENTS — FLAGGED BY ADEQUATE

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.

Wired AI

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

Tom's Hardware

Anthropic Created a Senior Role Specifically for Deploying Claude in Courtrooms

Artificial Lawyer

Anthropic loosens Fable 5's biology restrictions but keeps the guardrails on for virology and toxicology

The Decoder

DOGE Whistleblower Sues Elon Musk While Instagram Confirms Breach

Wired AI