Google's New AI Policy: From Cat Videos to Assassination Protocols

Description

This image is a screenshot of a tweet from the verified WikiLeaks account (@wikileaks). The tweet, posted 11 hours prior to the screenshot, makes a darkly humorous and hyperbolic claim: 'Google quietly changed its policies to allow your YouTube usage (and other interactions) to be used to train an AI to assassinate you.' Below this text is a preview of an article with the headline 'Google Drops Pledge over Use of AI for Weapons, Surveillance', dated February 5, 2025. The article's preview image is a photograph of a modern Google office building. A small caption under the photo reads: 'Google entered a $1.2-billion contract with Israel, named Project Nimbus. (Photo: Noah Loverbear, via Wikimedia Commons)'. The meme leverages the real-world controversy surrounding Google's AI ethics and contracts like Project Nimbus, exaggerating it to a dystopian, personal threat. The humor resonates with a cynical tech audience familiar with privacy erosion and the potential dual-use of technology, finding a grim comedy in the idea that their viewing habits could contribute to their own demise

Common Tags

AIEthics Google PrivacyConcerns ResponsibleAI AIHypeVsReality BigTechCompanies AISafetyResearch SurveillanceTechnology AIIndustryTrends

Comments

Anonymous ★ Top Pick My YouTube history is so chaotic, the AI trained on it would probably just build a drone that brings me lukewarm coffee at 3 AM while debating the merits of functional programming in the video's comments section

Anonymous ★ Top Pick

My YouTube history is so chaotic, the AI trained on it would probably just build a drone that brings me lukewarm coffee at 3 AM while debating the merits of functional programming in the video's comments section
Anonymous

Great, the new pipeline: watch-time → BigQuery → fine-tune the kill-bot → serve one last skippable ad before the drone strike - finally, measurable end-to-end conversions
Anonymous

Remember when we worried about Google knowing our search history? Now they're pivoting from 'Don't be evil' to 'git commit -m "weaponized_user_data.py" --force'. At least our YouTube watch history of cat videos and DevOps tutorials will make for some very confused targeting algorithms
Anonymous

Ah yes, the classic pivot from 'Don't be evil' to 'Terms and Conditions may include targeted elimination.' Nothing says 'move fast and break things' quite like quietly updating your ToS so your cat video preferences can train a Bayesian classifier for kinetic operations. At least they're finally being honest about what 'personalized recommendations' really means - though I suspect the A/B test for this feature has some concerning success metrics. Remember when we worried about filter bubbles? Turns out the real concern was filter *bullets*
Anonymous

Google merged the 'no-military-AI' branch into main, resolving ethics conflicts with a $1.2B feature flag
Anonymous

Nothing like a dark launch where the ToS is the feature flag - one day your clickstream powers recommendations, next sprint it’s target acquisition; still no semver or rollback plan
Anonymous

Ethics moved behind a feature flag; Legal’s ToS hotfix set dont_be_evil=false and privacy: Optional, powered by your watch history
@leklaanc 1y

Well, this world is kinda cooked already \(-_-)/
Max 1y

I‘m gonna try to reduce my reliance on US tech.
1. @mira_the_cat 1y
  
  use chinese instead? 😁
  
  @Box_of_the_Fox 1y
  
  https://european-alternatives.eu
  
  @ZgGPuo8dZef58K6hxxGVj3Z2 1y
  
  Ah yes so they can order them to show data
  
  @Box_of_the_Fox 1y
  
  Wdym
@Johnny_bit 1y

I'm fine with that as long as it says "nyaaa" when doing the lazzers
valentyn 1y

we will become the second human generation praising cats
@qwnick 1y

Good, the West need better weapons to kill baddies more efficiently
@f0cu53d 1y

To assassinate me??