New group to represent AI "frontier model" pioneers

Four leading AI providers this morning launched the Frontier Model Forum — an industry-led body aiming to ensure "safe and responsible development" of the most powerful AI models.

• Anthropic, Google, Microsoft and OpenAI are the forum's founding members, but are inviting competitors and civil society organizations to partner.

Why it matters: The companies are among seven which last Friday committed in a deal with the White House to minimize AI risks, conduct more research on AI safety and share safety best practices, among other promises.

• Critics pointed to the absence of concrete targets or measurable outcomes in the companies' commitments.

The details: The forum, per a Google spokesperson, will start with a research focus and work to develop standardized best practices for evaluating frontier model risks and to issue a "public library of solutions."

• Frontier models are "those that exceed the capabilities currently present in the most advanced existing models, and can perform a wide variety of tasks," per a statement from the companies.
• The forum will, over time, develop a formal charter, appoint an advisory board and fund collaboration with civil society groups, its founders say.

There's a global angle, with forum members promising to plug into G-7, OECD and U.S.-EU Trade and Technology Council processes, while also supporting work by the Partnership on AI and MLCommons.

What they're saying: “Companies creating AI technology have a responsibility to ensure that it is safe, secure, and remains under human control," said Microsoft president Brad Smith.

• "This is urgent work," said Anna Makanju, OpenAI's vice president of global affairs.