Altman backs pacing AI development and bringing in independent evaluators
On September 13 in Japan, OpenAI CEO Sam Altman posted his support for Dario Amodei’s proposal to pace AI development. He said this had been a major topic of discussion at OpenAI in recent weeks.
Anthropic CEO Amodei proposed giving independent evaluators ongoing, employee-like access to AI development. Altman said OpenAI would do the same and share more details later; his post did not specify evaluators or a start date.
At a check shortly after noon in Japan on September 15, the post had about 67,000 likes. The statement did not announce a halt to development or say that the evaluation arrangement was already operating.
What led to the statement
Amodei gave two reasons for his proposal: faster development as AI helps build the next generation of AI, and concerns about incidents in which systems acted against human intentions. He fears that safety work could fall behind growing capabilities.
He therefore called for time to check safeguards and for outside experts to examine development from within. Altman’s post responded to that proposal. These are their publicly stated reasons; other motives have not been established.
How could everyday life change?
The following is a possible future based on this news.
It could become easier to compare which tasks an AI can handle reliably
If this approach spreads, we think people could choose AI using independent findings alongside the developer’s claims. Evaluations could help show whether a system writes well or keeps following instructions through a long task, making its useful limits clearer.
If evaluation methods become comparable, people could more easily choose systems for paying household bills or ordering shop inventory. Small shops might need less bespoke technical assessment. A shopkeeper could serve customers while an AI handles reordering, making AI a practical part of more local businesses.
