Safety advocates push AI labs to cap R&D compute
The Information reports that a loose group of AI safety advocates and researchers is pushing AI companies to cap the compute they devote to research and development. The proposal would commit companies to a minimum share of compute for safety or monitoring and another minimum share for inference, limiting what remains for training stronger models. Separately, OpenAI CEO Sam Altman last week described three scenarios the industry has considered, including merging all AI labs into one entity or creating a global regulator modeled on the IAEA.
It adds a concrete compute-allocation mechanism to the wider debate over how labs can pace AI development so safeguards keep up with the technology.
This source provides an excerpt. Continue reading on the original site.