Google, OpenAI and Anthropic join forces on AI safety

Google, OpenAI and Anthropic are reportedly collaborating to establish an independent Artificial Intelligence safety body. It is expected that the standards will be developed by late 2026 or early 2027. The trio has reached out to former White House AI policy advisor Sriram Krishnan to serve as chief executive of the body. The report shared by The Information states that the planned organisation is provisionally named the Standards Authority For Frontier AI (SAFA). 

The Authority is said to operate independently of government administration with the aim to define practical benchmarks for three companies. The team will also provide direct support to developers to evaluate the cutting-edge models before they are launched for commercial use. 

As claimed in the report, the Authority has reached out to Arati Prabhakar, former director of the White House Office of Science and Technology Policy, to take the top post at SAFA. The tech giants are also weighing whether SAFA should conduct direct model assessment in-house.  

It is pertinent to note that while the United States Center for AI Standards and Innovation already handles federal evaluation, insiders noted concerns that the government agency lacks adequate technical and financial resources to thoroughly inspect complex frontier systems.

This comes as the trio remain locked in a race for commercial dominance. The momentum toward collective oversight accelerated earlier this month after Anthropic CEO Dario Amodei released an essay calling for an industry-wide slowdown in the release of frontier models to allow safety mechanisms to catch up.