Zuckerberg admits progress on Meta’s AI agent is falling short of expectations
During a recent internal meeting, Mark Zuckerberg, CEO of Meta, admitted that the development of AI agents – automated tools designed to perform tasks for users – was falling short of the company’s initial expectations. He noted that progress on these systems had not accelerated over the past four months as anticipated, meaning that the strategic goals linked to this technology have not yet been achieved.
Zuckerberg also reflected on the company’s sweeping reorganisation and acknowledged that the process had shortcomings. He admitted that management had misjudged the timing of the transition, resulting in a restructuring that went less smoothly than intended. Those changes, including the cutting of around 10% of the global workforce and the transfer in May of some 7,000 employees to AI roles, were intended to reallocate resources to expensive AI infrastructure and improve operational efficiency.
Despite these drastic measures, which have dented employee morale, Zuckerberg previously stated that no further large-scale job losses were planned for the current year.
The push towards this new structure was fuelled by concerns among top executives at the start of the year that Meta was moving too slowly to keep pace with the sector. Zuckerberg noted that management had been particularly optimistic during the planning phase in January and February about the capabilities of competing tools, such as Anthropic’s Claude Code.
Although Meta is investing heavily – expected spending on AI infrastructure this year amounts to €125 billion – the CEO believes the company will see a significant return on these investments within the next three to six months.
In addition, CTO Andrew Bosworth addressed a security breach involving software used to monitor employees’ digital activity and mouse movements. Following an investigation into the leaking of private data, Bosworth clarified that no employee data had been used for AI training.
Although the programme was temporarily suspended, Bosworth announced that any future deployment of this tracking software would be voluntary. This marks a shift from the initial rollout in April, when employees in the US were told they could not opt out of the monitoring programme.
Business AM
Subscribers 0
Fans 0
Followers 0
Followers