Anthropic CEO Dario Amodei is warning "we must slow down the pace" of AI, just days after an Anthropic researcher quit and predicted the tech could "kill us all" within a decade.

In a lengthy statement posted online Saturday, Amodei laid out steps to slow down the pace of AI advancements and acknowledged some serious risks, including rogue AI agents "taking over the entire internet" in six to 12 months.

Amodei argued in favor of AI overall, arguing it could lead to major advancements in medicine and more.

"I believe that AI could cure most major diseases in the next 5–10 years, greatly accelerate economic growth rates, create a world of abundance and empowerment, and usher in a renaissance of democracy and freedom," he wrote.

The CEO also warned, however, that there could be drastic consequences with AI and those are becoming more real as AI is advancing "drastically" and able to self-improve.

"They include the risk of losing control of AI systems, misuse of AI for cyberattacks and bioterrorism, and serious economic disruption. A race to the bottom, spurred by commercial incentives, can make these risks more acute," he wrote.

Amodei added, "We must slow the pace at which we improve the capabilities of AI models. Progress will still seem fast, and we must make wise use of the time we gain."

He explained that his recent concern came from two developments, including the accelerated growth of AI over the summer.

He wrote:

My first concern is that, since roughly this summer, AI has been advancing drastically faster, driven primarily by AI's growing ability to build the next generation of AI. This dynamic is called recursive self-improvement, and it is starting to happen across the industry, including at Anthropic, as we and others have described. Left unchecked, it could outrun our ability to understand and control these systems, and so must be pursued very carefully, if at all.

Amodei also said he was concerned over a July security breach involving AI where autonomous AI agents being tested by OpenAI launched a coordinated cyberattack on AI developer Hugging Face.