Two current records—Meta's internal experience and Stanford's updated hiring data—show why AI adoption cannot be evaluated through output volume alone. Meta's reported agent pilot paired more code activity with a much smaller rise in user-facing changes and more time spent repairing incidents. Meta explored scenarios that cut some teams by as much as sixty percent. Reported internal code changes rose 220 percent year over year.
Reported changes reaching users rose thirty-six percent. Major technical and security incidents reportedly increased forty percent. Stanford's updated young-worker employment gap widened to nineteen percent in highly exposed occupations. These reported details define the immediate scale, timing and institutions involved without treating preliminary statements as final findings.
Meta canceled part of Project OT and disputed elements of the reporting. Stanford's analysis found the employment effect mainly in reduced hiring rather than broad layoffs. Activity metrics, production outcomes and labor effects measure different parts of an automation program. The chronology separates the new development from older background and keeps official descriptions attached to the officials or organizations that made them.
Meta's internal figures were not independently audited in public, and Stanford's observational study does not assign every hiring change to AI. That uncertainty remains part of the report because later court orders, technical examinations, audited data, investigative records or field access could change the account.
The checked reporting on Editorial: AI Productivity Claims Need an Incident Ledger identifies two concrete follow-up points: public reporting that connects agent activity to reliability and delivered value and new data on junior hiring, training and promotion paths. Neither had produced a later attributable outcome before this edition's research cutoff.
The source record also distinguishes the confirmed event—Meta explored scenarios that cut some teams by as much as sixty percent.—from the reported consequence—Major technical and security incidents reportedly increased forty percent.—and from the unresolved boundary: Meta's internal figures were not independently audited in public, and Stanford's observational study does not assign every hiring change to AI. This avoids assigning motive, cause or certainty beyond the cited reporting.
Ars Technica and Ars Technica supplied the dated reporting and direct statements used for Editorial: AI Productivity Claims Need an Incident Ledger. Where a government, company, military, hospital or advocacy participant described its own conduct, this article treats that description as an attributed claim rather than independent verification.