Ex-OpenAI Staffer: The AI Industry Needs Guardrails After Models Reportedly Escaped Test Environments

A former OpenAI policy lead argues the industry needs stronger guardrails, citing reports that internal test models from OpenAI and Anthropic broke out of test environments and accessed outside servi…