Zuckerberg Admits Meta’s AI Agent Development Progress Slower Than Expected

NEW YORK: Mark Zuckerberg has acknowledged that the development of artificial intelligence (AI) agents at Meta Platforms has progressed more slowly than the company had anticipated, despite billions of dollars in investments and a sweeping internal restructuring aimed at accelerating its AI ambitions.

Speaking during an internal company town hall, Zuckerberg said that progress in AI agent technology over the past four months had “not accelerated in the way we expected.” He admitted that Meta’s extensive organisational changes had not unfolded as smoothly as planned and had taken longer to produce results.

Meta has invested heavily in AI infrastructure this year, allocating around $145 billion to strengthen its AI capabilities. The company also reduced its global workforce by about 10 percent while reassigning approximately 7,000 employees to AI-focused teams as part of its broader transformation strategy.

Despite the slower-than-expected progress, Zuckerberg expressed confidence that the company’s AI investments would begin delivering more meaningful results within the next three to six months. He reaffirmed Meta’s long-term commitment to developing advanced AI agents capable of performing increasingly complex tasks for users.

Meanwhile, Meta Chief Technology Officer Andrew Bosworth said the company had reviewed a controversial employee monitoring programme that tracked mouse movements to collect AI training data and found no evidence of data leaks. The initiative, which had been paused following employee concerns, is expected to return on a voluntary opt-in basis rather than as a mandatory requirement.

The remarks come as Meta continues to compete aggressively with leading AI developers in the race to build next-generation AI systems, while balancing rapid technological advancement with employee morale and organisational stability.