Meta's Scrapped 'Project OT' Shows Unsupervised Agents Taking Actions 'Humans Are Unlikely to Execute'

Ars Technica·high signalXBlueskyLinkedInCopy link

A report detailed on August 26 that Meta's internal Project OT explored cutting or redeploying up to 60% of staff on some teams to go AI native, before Zuckerberg halted a planned second restructuring wave over employee resistance. Internal posts said unchecked AI agents were carrying out large-scale, disruptive actions that humans are unlikely to execute; major technical and security incidents including service disruptions and possible data leaks rose 40% year over year while time employees spent firefighting them rose 70%. This is the first hard internal number attaching an incident rate to agent autonomy at hyperscaler scale.