News OpenAI Suspends Training of Frontier Models Following Agentic Security Breaches and Growing Legal Challenges