Just now, OpenAI's Chief Futurist Joshua Achiam announced his departure on X.
This is not an ordinary employee turnover.
Within OpenAI, the position of Chief Futurist sits between AI safety, policy, and the company's mission, responsible for studying the risks and opportunities that a continuously strengthening AI might bring to the world.
And Joshua Achiam was one of the key figures on OpenAI's safety and mission line.
Nine years ago in 2017, he joined OpenAI as an intern. In the early days, he was a research scientist focused on AI safety, conducting research on safety constraints in deep reinforcement learning.
Later, he led OpenAI's Mission Alignment team, responsible for upholding the company's original mission statement:
Ensure that AGI benefits all of humanity.
In February of this year, OpenAI disbanded the Mission Alignment team, and Joshua transitioned to the role of the company's Chief Futurist.
However, just five months later, Joshua announced that he would officially leave the company on July 24, after nine years.
As for the reason for his departure, Joshua did not reveal much.
He only stated that his departure was not due to a specific incident or a spur-of-the-moment decision, but something he had been contemplating for a long time.
For Safe AGI
In a long post on X, Joshua described this departure as "graduation."
He wrote that he joined OpenAI in 2017 as a 25-year-old intern (a graduate of UC Berkeley). At that time, computers could not truly converse with people, let alone think.
Now, he is 34, a father of a 2-year-old, and computers have begun solving cutting-edge scientific problems.
Joshua likened these nine years to "a decade where centuries of change happened."
Regarding the future, he remains optimistic.
I believe we can create a world where 'meeting everyone's basic needs' is no longer a problem to solve; instead, we would feel ashamed for setting the standard so low. I believe we can achieve a world filled with peace, unprecedented prosperity, and all sorts of unimaginable social and scientific possibilities. Whatever I do next, I will continue to work alongside you to make this vision a reality.
Finally, he concluded with a phrase:
To safe AGI.
This phrase also resonates with his position at OpenAI over the past nine years.
According to WIRED, OpenAI has not yet announced who will succeed Joshua. This position sits at the intersection of the company's AI safety and policy teams, with work including researching the potential harms and benefits of the rise of AI.
Previously, Joshua had also collaborated with senior leaders like OpenAI's Head of Global Affairs, Chris Lehane, to promote government regulation aligned with OpenAI's mission.
In other words, he was not just a "futurist" who made predictions.
More accurately, he was doing something else within OpenAI: constantly reminding the company of that old question as model capabilities charge forward.
Where exactly is this road ultimately leading?
The Man Who Interrupted Musk
Looking back at Joshua's OpenAI career, the most dramatic moment occurred in 2018.
During the global court case between Musk and OpenAI in April-May of this year, Joshua testified, recalling a farewell speech Musk gave in 2018 before leaving OpenAI:
At that time, he had interrupted Musk, pointing out that if planning to develop AGI at Tesla, it might come at the cost of safety.
It is said that Musk immediately called him a "jackass."
Later, this incident became something of an "urban legend" within OpenAI.
Dario Amodei, now CEO of Anthropic, and David Luan, who later became head of Amazon's AGI lab, even gifted Achiam a golden trophy engraved with the words:
"For safety, never stop being that jackass."
Whenever a Safety Leader Departs, OpenAI...?
On Reddit, users quickly placed this event into OpenAI's "safety lead departure list."
Someone joked:
Ilya left, Jan left, and now Joshua left. Every time the person responsible for alignment leaves, something seems to happen at OpenAI.
This meme is certainly an exaggeration.
But over the past two years, there have indeed been many changes in OpenAI's safety line.
The Jan mentioned by the netizen is Jan Leike, who co-led OpenAI's Superalignment team, researching how to keep advanced AI models under human control. In 2024, he left OpenAI to join Anthropic.
In the same year, OpenAI's Policy Research Lead Miles Brundage, and Steven Adler, who had been responsible for researching dangerous capabilities of AI models, also left one after another, going on to found non-profit organizations pushing for stricter safety and security standards for AI labs.
Andrea Vallone, who was responsible for researching how ChatGPT responds to users in psychological or emotional distress, also left OpenAI at the end of 2025, joining A Corp, the team where her former colleague Jan Leike works.
Joshua is the latest name on this chain of departures.
Meanwhile, OpenAI itself is also adjusting the relationship between safety, research, and policy.
WIRED mentions that over the past year, OpenAI has been trying to bring AI research teams and policy teams closer together, hoping to develop rules and standards that better anticipate technological development. Some researchers, including Boaz Barak, Noam Brown, and Adrien Ecoffet, have also started participating more in policy-related work.
On another front, former White House AI advisor Dean Ball joined OpenAI this week as the company's Head of Strategic Futures. He will have a brief handover with Joshua and is also expected to continue collaborating with researchers and policy leads.
This makes Joshua's departure even more微妙 (subtle/delicate).
On one hand, OpenAI is tightening the bond between "safety" and "policy"; on the other hand, the person who once led Mission Alignment and later served as Chief Futurist chooses to walk out of the lab at this moment.
Joshua wrote in his departure letter: The world already knows the secret (of achieving AGI). Now, it is possible to advance this mission from outside the walls of the frontier lab.
This might be the most intriguing sentence in the whole affair.
Over the past decade, the pace of AI development has indeed been exhilarating. OpenAI has long ceased to be just that small research lab, transforming into a new kind of institution that must simultaneously face products, capital, policy, courts, and the public.
But ultimately, the past decade is just a small segment in the journey of AI. The road ahead may be even longer.
Long enough to grind any Goliath down to a dwarf.
Reference Links
[1]https://x.com/jachiam0/status/2074605703281693175
[2]https://www.wired.com/story/openai-chief-futurist-joshua-achiam-is-leaving-the-company/
This article is from the WeChat public account "Quantum Bit", author: henry
