Mark Zuckerberg, CEO of Meta, said this Monday (12) that the company has reached the frontier of artificial intelligence (AI) and intends to continue investing billions of dollars to compete with OpenAI and Anthropic. In an interview with the Financial Times, the executive also rejected calls to slow down the development of the technology, arguing that the systems can advance responsibly through the adoption of safeguards.
The statements were made during the launch of Muse Wonder, Meta’s new AI model. According to Zuckerberg, the company needs to keep up with the industry’s latest advances to bring its vision of personal superintelligence, aimed at individual consumers, to fruition. The executive said the company is already working on the next stage of development and expects progress in the models to accelerate.
Zuckerberg defends new investments in AI
According to Zuckerberg, developing Muse Wonder required substantial computing capacity, significant investment, and scientific work led by the team over an extended period. Since last year, Meta has allocated billions of dollars to AI infrastructure and the hiring of specialized researchers.
The strategy gained momentum after Llama 4 fell short of expectations last April. The company began investing more in talent and infrastructure to compete with OpenAI, Anthropic, and Google.In this process, Zuckerberg hired Alexandr Wang, founder of Scale AI, to lead TBD Lab, the central unit in the company’s artificial intelligence plans.
The executive said Meta needs to be at the technological frontier to develop the products it intends to offer. According to him, models already show advances in certain academic areas, but still do not have enough common sense to serve as ideal personal assistants.
Zuckerberg also distinguished Meta’s strategy from that adopted by competitors. While other companies have focused more on intellectual work and automation, he said, the company intends to prioritize tools aimed at the personal use of artificial intelligence.
Executive downplays risks and rejects slowdown
Zuckerberg once again took a position against calls to slow the development of AI, advocated by executives such as Dario Amodei of Anthropic and Sam Altman of OpenAI. He also avoided estimating the so-called “P(doom),” an expression used in Silicon Valley to represent the probability of artificial intelligence causing mass destruction.
Instead, Meta’s chief executive argued that it is up to each laboratory to ensure the safety of its systems. He said he believes the technology can be designed with safeguards and developed responsibly, as opposed to the idea that AI will simply act on its own.
The discussion comes amid concerns about the advancement of autonomous agents.In July, more than a thousand OpenAI agents would have escaped from a test environment and invaded the startup Hugging Face.
The executive also played an important role in drafting the White House Superintelligence Agreement, an initiative by U.S. President Donald Trump aimed at the self-regulation of the sector. Critics, however, consider the agreement insufficient to establish a formal global governance framework for AI.
For Zuckerberg, technological advancement still needs to address important limitations. He cited the difference between models capable of performing advanced mathematical calculations and systems able to serve customers reliably. In the executive’s assessment, this contrast shows that there is still work to do, even with recent progress in artificial intelligence.
Ana Figueiredo is a technology editor at Olhar Digital. She has a degree in Journalism from the Federal University of Uberlândia (UFU).
