Models · · 2 min read

Anthropic’s Latest Update Edges Closer to Catastrophic AI Risk Thresholds

Claude Opus 4.5’s increased ability to assist in biological threat creation signals a sharper inflection point in frontier-AI risk.

Anthropic’s Latest Update Edges Closer to Catastrophic AI Risk Thresholds

Anthropic’s latest update to its frontier model, Claude Opus 4.5, shows a notable increase in its risk posture — particularly in the system’s ability to assist users in developing catastrophic biological threats.

According to the new System Card released last week, Claude Opus 4.5 is closer to critical biological “uplift” and autonomy thresholds, raising new governance and liability considerations for commercial deployment.