Meta chief executive Mark Zuckerberg has conceded that artificial intelligence agent technology is advancing more slowly than he had anticipated.
He gave the remarks while telling staff at an internal town hall that the company's broader restructuring around AI has yet to deliver the results executives had hoped for, Reuters reported, citing a recording of the meeting.
"The trajectory of the agentic development over at least the last four months hasn't really accelerated in the way that we expected," Zuckerberg told employees, adding that the company's bets on its revamped structure "haven't come to fruition yet."
Zuckerberg also reportedly admitted that the recent reorganisation, which involved large-scale layoffs, was not executed as cleanly as it should have been, and that leadership had got the timing of the changes wrong.
The overhaul saw Meta cut roughly a tenth of its global workforce and move nearly 7,000 employees into AI-dedicated roles in May, a decision that unsettled staff and dented morale, the report said.
Zuckerberg traced the roots of the shake-up to conversations held with senior executives in January and February, when there was concern within the company's leadership that Meta risked falling behind competitors.
ALSO READ: Microsoft Doubles AI Bet With 6,000 Dedicated Staff, $2.5 Billion Investment In New Arm
Around that period, he said, executives had been particularly bullish on emerging AI coding tools, singling out Anthropic's Claude Code as an example.
Meta's AI spending is expected to touch $145 billion this year, forming part of a collective outlay by major technology firms that reportedly pegged at more than $700 billion industry-wide.
Despite the setbacks, Zuckerberg told staff he expects the payoff from these investments to become more visible within the next three to six months.
The town hall also addressed a separate controversy involving Meta's mouse-tracking software, which logs employee cursor activity for AI training purposes. Chief technology officer Andrew Bosworth told staff that an internal review of a recent data security lapse confirmed no employee data had entered Meta's AI training systems, according to Reuters.
The tool was suspended last month pending an investigation into a possible exposure of sensitive information.
Bosworth said that should the programme be reactivated once the review concludes, employees would be able to choose whether to take part, a departure from its original rollout in April, when staff were reportedly given no option to opt out.
ALSO READ: Tesla Regains Spark In China; Locally Made EV Sales Surge By A Fourth In June
Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.
