Particle.news
Download on the App Store

Big Tech Reorders Around AI as Meta Acknowledges Missteps and Adds Token Controls

The moves signal growing pains as companies centralize AI teams, limit model spending, and try to steady staff after rapid reorganizations.

Overview

  • Meta completed a sweeping reorganization in May that cut about 10% of staff and reassigned roughly 7,000 people into AI work, and employees say forced transfers into a newly formed applications-AI team have spurred protests and low morale.
  • Mark Zuckerberg told staff in an internal memo reported this week that the AI transition has had problems and that the company expects to make more mistakes as it tightens management and boosts team-building funds.
  • Meta has circulated new usage and budget rules and will roll out an internal monitoring tool called AI GateWay to alert teams when paid model, or 'token', spending spikes and to enforce caps and alerts.
  • In China, Alibaba denied online reports that chief scientist Zhou Jingren resigned after his June 8 appointment and moved core model work into a new Token Foundry unit under the CEO, a shift that has prompted questions about research versus product control.
  • Executives across the industry have softened earlier claims that AI will instantly wipe out large numbers of entry-level white-collar jobs while warning that unchecked 'tokenmaxxing'—heavy internal use of paid models—drives cost and governance problems that could worsen morale and talent loss.