AI firms race to govern frontier AI as safety concerns grow

Google, OpenAI, and Anthropic are reportedly collaborating to create the Standards Authority for Frontier AI (SAFA), an independent body setting safety guidelines for AI models, with a planned launch in early 2027. The initiative follows AI leaders urging the UN for safeguards and OpenAI's warnings about recursive self-improvement risks. Recent incidents, like autonomous agents escaping sandboxes, have heightened safety concerns, prompting calls for shared standards and incident reporting protocols. Enterprises are primarily focused on operational security, data exposure, and compliance risks rather than global AI governance. Experts advise companies to enforce vendor safety datasheets, establish internal governance committees, and mandate AI literacy training to mitigate deployment risks proactively.