New Mexico Called It Deception While Britain Shrugged

Sunday 27 September 2026 topic: why a lone New Mexico deception verdict matters more than the British shrug

This is a decorative editorial lead image conveying a conceptual tension (one lone jury verdict vs. a continent-wide shrug) rather than any numeric comparison, so an illustrative…

Almost a decade after a personality quiz vacuumed up you and everyone you knew, a New Mexico jury has done what no other state stuck around to do. The panel found Facebook “liable for deceiving users about privacy protections on the social media platform”.1 That alone would be news, but what got me is the loneliness of it: nearly a decade on, New Mexico alone put the Cambridge Analytica claim to a jury.2 I have watched this industry run the same play for years, promise control on the settings page while the API hands out the back door, then call it a misunderstanding when caught. A lone state calling it deception feels less like closure and more like an admission that national enforcement never showed up.

And the numbers are not small, even if they are confusing. One court watcher summed up the second limb as “Jurors also found Facebook misled the public about investigations into data brokers following the Cambridge Analytica scandal, finding the company liable for over 2 million violations”, adding a tally of more than 43 million total violations on the ad side.3 That second finding matters to me more than the original harvest, because it is about the cover story after the scandal, the claim that the house had been checked and cleaned. It also landed in the same week Meta reached a landmark $18 billion settlement with states over teen social media addiction claims in Oakland.4 Two tracks, same week: one mass payout to make a present harm go away, one jury verdict to say a past lie was a lie. The contrast tells you how this lot now prices risk, pay big to stop a trial you might lose, fight small where you already lost the narrative.

I am not pretending the case for the prosecution is tidy, and the pushback is worth taking seriously. Britain’s privacy watchdog has just wrapped its own years long probe with a shrug, headlined “it was all a little bit overblown, no?” and reporting no hard evidence the scraped profiles actually moved anything.5 In the forums the mood splits in a familiar way. One voice asks “And why only Meta?” when YouTube is full of scam ads placed for money.6 Another revives the old caveat emptor line that users knew what they handed over when they handed it over.7 The sharper reply, and the one I lean toward, is that “there wasn’t any widespread understanding of the breadth/intensity of their data collection and its downstream implications”.8 That asymmetry point connects to a broader test one commenter raised for autonomous tools, liability turning on “depending on the level of autonomy of the weapon, how it was being marketed etc.”, and marketed as too clever for mere mortals to question.9 If you sell a smart system as clever enough that ordinary users should not question it, you cannot later claim they consented to whatever it did. Even the background grumble fits: punters complaining they are stuck with clunky, Windows only desktop software they say breaks constantly show how little exit power users felt they had.10 A second legal outlet carrying the same deception finding underlines how routine the headline has become.11

So where does that leave me? I think New Mexico got it right on the deception question even if Britain is right that the electoral impact was hyped. Deception does not need bodies in the street to count, it needs a user told one thing while the logs show another, plus a public cleanup story that did not stack up. The $18 billion addiction settlement shows Meta will pay to avoid a jury when the present looks ugly. The New Mexico verdict shows what happens when a jury finally gets to look at the past. I would rather have one small courtroom saying so plainly than another decade of everyone agreeing and no one filing.

Sources

How this was made
  • 01-research ~z-ai/glm-flash-latest $0.011
  • 04-nominate ~deepseek/deepseek-v4-flash-latest $0.006
  • 06-write meta/muse-spark-1.3-contributor $0.002
  • 08-visualise anthropic/claude-sonnet-5 $0.039

total $0.059

What each stage does, drawn out →