Zuckerberg says Meta's AI agent push is moving slower than expected

WASHINGTON: Meta chief executive Mark Zuckerberg told employees the company’s push into AI agents has not advanced at the pace management expected, according to a recording of an internal town hall heard by Reuters and reported by Express Tribune.

Zuckerberg also acknowledged problems in the company’s broader restructuring, saying the overhaul was not carried out as smoothly as it could have been and that executives had misjudged the timing of the changes. The social media company has been trying to ease some of the internal disruptions caused by the reorganisation introduced earlier this year while keeping its overall direction intact.

Meta cut about 10% of its global workforce in May and reassigned around 7,000 employees to teams focused on artificial intelligence. The measures drew employee criticism and added to concerns about morale. The restructuring was part of a wider effort to support expensive investment in AI infrastructure and to place the company in a stronger position to benefit from efficiency gains linked to AI-assisted work.

Zuckerberg had told staff in May that he did not expect another round of companywide layoffs this year, although some employees remained doubtful. Looking back on recent progress, he said the expected acceleration in agentic AI development had not materialised over the past several months and that the company’s assumptions behind the new structure had yet to deliver the results it anticipated.

He said discussions with senior company figures in January and February, when planning for the restructuring was underway, had centred on concerns that Meta might not adapt quickly enough. At that stage, he said executives were highly optimistic about tools such as Claude Code from AI startup Anthropic.

Meta is projected to spend as much as $145 billion on AI infrastructure this year, part of a broader Big Tech outlay of more than $700 billion on the technology. Zuckerberg told employees he expects the company to begin seeing more meaningful returns from those AI investments within the next three to six months.

A Meta spokesperson declined to comment on Thursday.

Review of mouse-tracking software

During the same town hall, Meta chief technology officer Andrew Bosworth discussed a review of a recent data security incident involving the company’s mouse-tracking software, which has drawn controversy. According to Bosworth, the review found that no employee data was used in AI training.

Last month, Meta suspended the programme while it investigated the exposure of sensitive information. The software tracks employees’ mouse movements and other digital activity for AI training purposes.

Addressing what could happen after the review is completed, Bosworth said that if the programme is reinstated, it would operate on an opt-in basis.

He told employees “For people who are comfortable, that's great, they can contribute to this kind of great human survey. To people who are not, it is not an issue”.