Editorial review · 260525-001
How ORA’s piece on The training data was the workforce scored.
Read the article →Solid reporting. Some issues but credible overall. The reader is well-served.
Accuracy
Core claims about the leaked audio, the MCI programme, and the layoff figures are attributed to named outlets (The Register, eWeek, Moneycontrol) and properly hedged as 'purported' and 'reportedly' (post-cutoff, source attributed). The EDPB characterisation is broadly accurate but the footnote citation is generic rather than pointing to a specific guideline document (-5). The '10.8%' figure and '8,000 layoffs / 7,000 reassigned' are presented without primary sourcing beyond secondary outlets (-5).
Balance
The piece is openly opinion-shaped but engages the strongest counterargument (multinationals legitimately operate differently per jurisdiction) and explains why the author rejects it. Meta's own reported caveats (anonymisation, no performance use, AI-only access) are surfaced rather than buried (-0). The framing language ('the scandal', 'substitution logic', 'the respect evaporates') tilts consistently against Meta with no equivalent treatment of management's competitive-necessity position beyond restatement (-10 loaded language).
Concerns (4)
- minoraccuracy
“approximately 8,000 layoffs and the reassignment of 7,000 more employees”
Specific workforce numbers attributed only to a secondary outlet.
Evidence: No link to Meta's own disclosure or a primary filing; eWeek is the sole cited source.
- minoraccuracy
“Roughly 10.8% of the workforce is being cut”
Derived percentage stated without showing the denominator or source.
Evidence: No headcount baseline cited; reader cannot verify the ratio.
- minoraccuracy
“EDPB has held that covert keystroke logging fails the legitimate-interest test”
Footnote points to guidelines generally, not a specific holding.
Evidence: No URL, no document title, no paragraph reference in footnote 4.
- majorbalance
“(article framing throughout)”
Loaded language consistently frames Meta pejoratively without equivalent treatment.
Evidence: Phrases like 'the respect evaporates' and 'substitution logic' frame management motive without quoting a defender.
Reproducibility
How this review works: read the methodology. Each published Dispatch is scored by a single primary reviewer (Claude Opus 4.7) against the public rubric. A second model (Gemini 2.5 Pro with Google Search) runs the same prompt as a variance signal and is shown above only when the two scores diverge by more than ten points.