Goblin House
Claim investigated: The absence of Project Maven references in earlier 2019 SEC filings suggests either new contractor onboarding or program classification changes that triggered first-time disclosure requirements Entity: Project Maven Original confidence: inferential Result: WEAKENED → INFERENTIAL
The claim faces a fundamental data integrity problem: the established facts state Project Maven itself filed SEC documents (facts 14-15), which is impossible since it's a Pentagon program, not a corporate entity. However, the core inference about new contractor onboarding or classification changes remains plausible given the documented 15-month gap between Google's withdrawal (June 2018) and alleged Q3/Q4 2019 disclosures, which aligns with typical defense procurement timelines.
Reasoning: The contradictory primary facts (Project Maven filing its own SEC documents) undermine the evidentiary foundation. While the timing analysis supports contractor transition scenarios, the data corruption prevents elevation to secondary confidence until the actual filing entities and contexts are identified.
SEC EDGAR: Project Maven in 10-K and 10-Q filings for Palantir, Alphabet/Google, Booz Allen Hamilton, CACI, SAIC between September-December 2019
Would identify the actual corporate entities making Maven-related disclosures and the specific contexts (risk factors, material contracts, or revenue recognition)
USASpending: All AI, machine learning, and computer vision contracts to Palantir Technologies between June 2018-December 2019
Would reveal whether Palantir received Maven-related contracts under different program names or procurement vehicles
SEC EDGAR: EDGAR accession numbers and full text for any filings mentioning 'Project Maven' filed on 2019-09-27 and 2019-12-20
Would resolve the data integrity issue by identifying the actual filing companies and whether Maven was mentioned in risk factors or contract disclosures
other: Pentagon JAIC contract awards and Other Transaction Authority agreements 2018-2019 via FOIA
Would identify alternative procurement mechanisms used for Maven and related AI programs that bypass standard USASpending reporting
SIGNIFICANT — This finding reveals either systematic data integrity problems in defense AI contract tracking or deliberate obfuscation of Project Maven's corporate participants through alternative procurement mechanisms, both of which have implications for oversight of military AI development.