Google, OpenAI, Anthropic plan joint AI safety body

Google, OpenAI and Anthropic are working together to establish an independent organisation aimed at strengthening the safety of artificial intelligence, with the body potentially beginning operations by the end of this year or early 2027, and the companies reportedly approaching former White House AI policy adviser Sriram Krishnan about serving as its chief executive.

Tentatively named the Standards Authority for Frontier AI (SAFA), the proposed organisation would operate independently of government oversight, reviewing AI safety commitments made by the three companies while developing new industry standards. It could also assist developers in assessing the safety of advanced models before their commercial release, although discussions are still underway over whether SAFA would confine itself to setting standards or would also conduct model safety assessments directly.

Read More: Bill Gates urges tighter oversight of powerful AI

The initiative comes amid intensifying competition among the three companies for dominance in the AI market, alongside growing scrutiny over the safety and cybersecurity risks posed by increasingly capable models. In July, moreover, Google DeepMind co-founder and Alphabet Chief Scientist Demis Hassabis had already proposed creating a US-based AI standards body, either through a public-private partnership or as a self-regulatory organisation for the industry.