AI Governance

• OpenAI asks California to tighten frontier AI safety law

  OpenAI wants California’s SB 53 expanded to cover frontier-model monitoring during training and evaluation plus stronger lifecycle cybersecurity protections.

• OpenAI slows frontier model training as cyber risks grow

  OpenAI has paused parts of frontier model training and tightened security controls as stronger cyber capabilities expose new risks in AI research infrastructure.

• Brazil launches R$2.3bn sovereign AI infrastructure push across rival technology ecosystems

  Brazil is funding major AI infrastructure in Rio de Janeiro and Rio Grande do Norte, raising strategic questions about sovereign compute, suppliers and governance.

• OpenAI Tests Cross-Interaction AI Safety Without Retaining Customer Data

  OpenAI previews Private Safety Processing to detect cross-interaction misuse while preserving Zero Data Retention for eligible API customers.

• Anthropic Expands Claude Mythos 5 Defense Without Opening Direct Access

  Anthropic brings Mythos 5 to Claude Security, funds open-source defense and expands verified access while keeping direct model use restricted.