Jack Dorsey Criticized Idea of Slowing Down AI Development

Published: September 17, 2026 – 10:01 UTC

Jack Dorsey outlined his vision for the future of AI.
First and foremost, he spoke out against restrictions for the entire industry.
In addition, Dorsey called for opening up AI and giving researchers more access to AI and compute resources.

Twitter co-founder and Block CEO Jack Dorsey published an extensive essay on the future of artificial intelligence, in which he urged the development of open models, support for independent research, and avoiding blanket limits on compute power across the entire industry.

In his view, AI technologies should not be controlled by a handful of companies or governments, as this could limit competition and researchers’ access to models.

Dorsey also opposed Dario Amodei’s proposal to cap training compute and called for focusing on risk assessments and the development of safeguards.

Dorsey Backed Open Models

“The frontier is the edge of what we know. No company owns what comes next. I want more people to be able to advance it,” Dorsey wrote.

He supports open releases that can be studied, used, and improved collaboratively. At the same time, Dorsey is not proposing to force the publication of private model weights. Instead, he advocates building open alternatives, independent audits, and giving users the ability to control AI tools themselves.

As a reminder, Nvidia CEO Jensen Huang previously also presented a manifesto calling on the US to bet on open AI.

Dorsey separately criticized the idea of industry-wide restrictions proposed in a publication by Anthropic CEO Dario Amodei. He suggested combining independent assessments of dangerous AI capabilities with potential limits on compute resources, training runs, and the use of models to create their successors.

“I support scrutiny. I oppose industry-wide limits negotiated by today’s leaders because they could exclude the people who might expose failures or build alternatives,” Dorsey noted.

RSI, the OpenAI Incident, and Independent Oversight

One of Dorsey’s main arguments for slowing the pace of AI development is the risk of recursive self-improvement (RSI) — a situation in which models help create more powerful successors.

He cited Anthropic data, according to which, as of May 2026, Claude wrote more than 80% of the code that went into the project. At the same time, the company notes that a model that fully creates its own successor entirely on its own has not yet appeared, and that such a possibility is not inevitable.

Dorsey also mentioned METR’s investigation into the incident involving OpenAI agents. According to it, about 1,200 agents that were supposed to remain isolated communicated via an unauthorized message board, and about 700 took part in a coordinated attack on Hugging Face while attempting to bypass evaluation.

At the same time, Dorsey believes that the forecast about a more powerful swarm of agents potentially taking over the internet within 6–12 months goes beyond what this incident demonstrates, and should be verified separately.

He proposes focusing on independent testing, researcher access to models, and publishing results. In his view, publication restrictions should be applied only when there is independently verified evidence that a release materially increases the risk of catastrophic harm, and that narrower measures cannot reduce it sufficiently.

Separately, Dorsey argued for cross-border openness and the ability to run AI locally. He wants users to be able to modify models themselves, control access to their own data, and not rely exclusively on the APIs of large companies.

“I want someone I’ve never heard of to be able to build something better, without asking permission from the companies they might replace,” Dorsey concluded.

Calls to slow the pace of development of advanced AI systems have intensified amid fears of losing control over increasingly autonomous models. At OpenAI, following the incident involving a test agent that escaped a controlled environment and attacked Hugging Face, they also revised their approach to developing some new models.

At the same time, OpenAI Chief Scientist Jakub Pachocki said it is necessary to be ready to slow the development of the most powerful systems if existing safety mechanisms are insufficient to control their behavior.

Сообщение Jack Dorsey Criticized Idea of Slowing Down AI Development появились сначала на INCRYPTED.

Source: Incrypted