DECLASSIFIED · RELEASED FOR PUBLIC REVIEW
OSA OFFICE OF SITUATIONAL AWARENESSRECORDS DIVISION · CITIZENBANNED.COM
FILE № OSA-181 CONFIRMED

CAMBRIDGE ANALYTICA

270,000 people took a quiz, and the data of 87 million was collected
HARVESTEDData on up to 87 million Facebook users, from about 270,000 app installs
MECHANISMFacebook's Graph API, which then exposed friends' data to app developers
EXPOSEDMarch 2018, by Christopher Wylie via The Guardian and Observer, and Channel 4
STATUSConfirmed as to harvesting. Effectiveness of the targeting: unproven

THE CLAIM

That a political consultancy improperly obtained personal data on tens of millions of Facebook users and used it to psychologically target voters in the Brexit referendum and the 2016 US election.

THE THEORY, AS ITS PROPONENTS TELL IT

This file has a well-evidenced half and an oversold half, and separating them is the point of including it.

The harvesting is established. Researcher Aleksandr Kogan built a personality quiz app. Around 270,000 people installed it, and Facebook's platform at the time allowed apps to collect data not only on users but on their friends. The resulting dataset covered up to 87 million people, the overwhelming majority of whom had never heard of the app.

Kogan passed the data to Cambridge Analytica, which breached Facebook's terms. Facebook learned of this in 2015, asked for deletion, accepted assurances, and did not verify or notify users.

The oversold half is psychographic targeting. The company's own executives, recorded by Channel 4 undercover, boasted about bribery, honey traps and swinging elections. Its public pitch claimed personality-based microtargeting could shift voters decisively. Academic assessment of psychographic targeting since has found effects that are real but small, and several campaign staff who used the firm's products have described them as unremarkable.

WHAT IS KNOWN

The data collection and the transfer to Cambridge Analytica are established and were admitted by Facebook.

Facebook was fined $5 billion by the Federal Trade Commission in 2019 and £500,000 by the UK Information Commissioner, the maximum available under the older UK regime.

Cambridge Analytica and its parent SCL Group entered administration in May 2018.

The UK Information Commissioner's final report in 2020 found the firm's actual capability was far less sophisticated than its marketing claimed, and found no evidence it had been used successfully in the Brexit referendum.

EVIDENCE FOR

  • Facebook's own admission of the data exposure and the scale involved.
  • Whistleblower documentation from Christopher Wylie, including contracts and data samples.
  • Undercover recordings of executives describing improper campaign methods.
  • Regulatory findings and record fines in two jurisdictions.

EVIDENCE AGAINST (THE EFFECTIVENESS CLAIM)

  • The UK regulator's own investigation concluded the firm's models were less advanced than advertised.
  • Peer-reviewed work on psychographic political targeting finds effects that are measurable and small, not decisive.
  • The firm had a strong commercial incentive to overstate its powers, and its most dramatic claims come from its own sales pitch and from executives caught boasting.

ASSESSMENT

Confirmed as a data scandal and unproven as a mind-control operation, and the second half matters as much as the first. Something genuinely improper happened: the data of tens of millions of people was taken through a permission none of them gave, and the platform that allowed it did not tell them. What did not happen, on the regulator's own finding, is the decisive psychological manipulation of two elections. This archive is as interested in that distinction as in the breach, because the exaggerated version is the one everybody remembers.

WHAT WOULD CHANGE THE GRADE

  • CONFIRMED as a data scandal: up to 87 million people's data collected from about 270,000 app installs, through a Facebook platform permission none of the friends gave. Facebook admitted it, was fined $5 billion by the FTC and £500,000 in the UK, and had accepted deletion assurances in 2015 without verifying or notifying users.
  • NOT established, and this is the half everyone remembers: decisive psychological manipulation of elections. The UK Information Commissioner's 2020 final report found the firm's models far less sophisticated than its marketing, and found no evidence of successful use in the Brexit referendum.
  • Peer-reviewed work on psychographic political targeting finds effects that are real and small. The most dramatic claims originate in the company's own sales pitch and in executives caught boasting on camera.

FURTHER READING

RELATED FILES IN THIS ARCHIVE

OSA RECORDS DIVISION RELATED: OSA-016 PRISM & THE SNOWDEN DISCLOSURES REV. 2026-08

The Office of Situational Awareness is an independent publication. Evidence grades are editorial assessments of the public record, not statements of new fact. Corrections welcome.