Breaking

Google, OpenAI and Anthropic Move to Build Independent Frontier AI Safety Standards Body

September 24, 2026 at 12:51 EDT

Policy & Regulation
Foundation Models
Security

Google, OpenAI and Anthropic are advancing a plan to create an independent frontier AI safety standards body tentatively called the Standards Authority for Frontier AI, or SAFA, with a target launch by late 2026 or early 2027. Under the plan, the body would support third-party testing before advanced model releases and define how developers should report safety and security incidents.

Sept. 24, 2026 · Google, OpenAI, Anthropic

SAFA pushes frontier AI safety toward audits

Three leading AI labs are advancing an independent standards body meant to support predeployment testing and common incident reporting for frontier models.

labs driving the proposal

current FMF member firms

>$10M

FMF safety research funding

Late 2026
or early 2027

target launch window

Three-lab start, six-member precedent

The proposed SAFA effort is narrower than the existing Frontier Model Forum membership.

SAFA backers

FMF members

Takeaway

Credibility will depend on whether the new body expands participation beyond its initial three-lab center.

What the standards body would do

Set safety standards

→

Support third-party tests

→

Standardize incident reports

Why supporters like it

Auditable procedures could turn safety commitments into repeatable checks before release.

Why critics are wary

Rules shaped by the largest labs could raise entry barriers unless governance is broad and independent.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.