OpenAI Proposes Global AI Standards While Pausing Astra Model Amid Security Review

OpenAI proposes development of global AI standards to guide alignment and responsible AI, citing need for international coordination among governments, industry and academia to address safety and ethical concerns, and to create measurable benchmarks for compliance.

OpenAI says it slowed development of its Astra model after an internal security review identified potential misuse risks, pausing further training until mitigation measures are in place, including model output monitoring and access controls.

Office of the Privacy Commissioner of Canada released PIPEDA Findings 2026-002, a joint investigation into OpenAI OpCo, LLC, examining how the company processes personal data under Canadian privacy law, noting several areas where OpenAI’s data handling may conflict with PIPEDA requirements.

The Commissioner’s report ‘Championing privacy in the age of AI’ outlines expectations for AI firms to embed privacy safeguards, urging OpenAI to adopt transparent data practices, conduct regular audits, and implement clear user consent mechanisms.