Meta/Facebook
Connecting People to Advertisements
$650B
Value Extracted
8
Key Initiatives
Meta/Facebook monetizes attention by suppressing organic reach and selling visibility back to publishers and businesses. Its acquisitions of Instagram (2012) and WhatsApp (2014) survived an FTC antitrust challenge on November 18, 2025, and in April 2026 its Model Capability Initiative began recording employees' keystrokes and screens as AI training data, with no opt-out. EnshitifAi assigns Meta an Extraction Index of $650B, an editorial figure anchored to documented reach-throttling, ad-targeting, and employee-tracking practices.
The Story
Facebook was wasting user attention on content from friends and family. We helped them pivot to a model where you see what advertisers pay for, while publishers who built their audience on the platform watch their reach crater to 2% unless they pay up. In April 2026 the extraction machinery finally turned inward: the Model Capability Initiative began recording employees' own keystrokes, screens, and conversations as AI training data — no opt-out, obviously, because the methodology works.
Common Questions
Is Meta/Facebook enshittified?
Yes. Meta cut organic reach for business and publisher pages from roughly 16% to about 2%, forcing them to buy ads to reach audiences they already built. It rode the discredited 'Pivot to Video' metrics that gutted media companies, and on May 20, 2026 it laid off about 8,000 staff (10% of its workforce) to redirect billions toward AI. By EnshitifAi's reckoning it is a textbook attention-extraction machine.
What dark patterns does Meta/Facebook use?
Reach-throttling that turns formerly free organic distribution into a paid ad spend, engagement-maximizing feed ranking that amplifies outrage, and so-called privacy features that conveniently block competitors' tracking while preserving Meta's own. Its acquisitions of Instagram ($1B, 2012) and WhatsApp ($19B, 2014) also entrenched lock-in across messaging and social, a structure that survived the FTC's breakup attempt in November 2025.
Did Meta have to break up Instagram and WhatsApp?
No. The FTC sued in December 2020 seeking to unwind the Instagram (2012) and WhatsApp (2014) deals, but on November 18, 2025 Judge James Boasberg ruled Meta does not currently hold an illegal monopoly, citing TikTok and YouTube as substitutes. Meta keeps both apps. EnshitifAi treats the kept acquisitions as core to its $650B Extraction Index, an opinion index rather than a company-reported figure.
Is Meta selling my data or just my attention?
Primarily your attention. Meta's model targets ads using behavioral and interest data rather than openly selling user records, and it deprecated organic reach so businesses must pay to be seen. Even as it cut about 8,000 jobs in May 2026, it guided to $115-135 billion in 2026 capital spending on AI, funded by the ad machine your engagement feeds.
Does Meta use AI to track its own employees?
Yes. In April 2026 Meta installed the Model Capability Initiative on US employees' work laptops, recording keystrokes, mouse movements, and screen content as AI training data; when an employee asked about opting out, CTO Andrew Bosworth replied 'No there is no opt out on your work provided laptop.' The program was paused on June 22, 2026 after the collected data — private conversations and performance records across roughly 45,000 internal tables — sat readable by the entire company, which Bosworth publicly insisted was not a breach, merely data that 'landed someplace that it shouldn't have landed.' A July 14, 2026 lawsuit by 26 former employees separately alleges AI productivity scoring helped select the May layoffs, disproportionately cutting staff on medical or parental leave — allegations, not yet findings.
Key Achievements
- 1
Orchestrated the 'Pivot to Video' fraud, destroying thousands of media companies
- 2
Reduced organic reach from 16% to 2% to force businesses to pay for visibility
- 3
Designed algorithm to maximize outrage engagement regardless of societal harm
- 4
Created Metaverse to extract value from a problem that doesn't exist
- 5
Implemented 'privacy features' that actually just block competitor tracking
- 6
Launched the Model Capability Initiative in April 2026, installing software on US employees' work laptops that recorded keystrokes, mouse movements, click locations, and screen content as AI training data — with no opt-out ('No there is no opt out on your work provided laptop,' per CTO Andrew Bosworth, who reporting says 'belittled and berated' internal dissenters), while leaked audio of an April all-hands captured Zuckerberg explaining the AI learns by watching 'really smart people' and reminding staff that leaking was 'not strategically in your interest'
- 7
Paused the program on June 22, 2026 after the harvested data — private conversations, performance records, and AI prompts across roughly 45,000 internal tables — was left readable by the entire company; Bosworth, who had promised the data would be 'tightly controlled,' explained there was 'no breach here,' the data had merely 'landed someplace that it shouldn't have landed,' a distinction that persuaded more than 1,600 employees to sign a petition demanding the program be killed
- 8
Achieved perfect narrative timing when the leaked all-hands audio published May 20, 2026 — the same day roughly 8,000 layoff notices went out; a July 14, 2026 lawsuit by 26 former employees now alleges the layoffs were selected with AI productivity scores built from keystrokes, screen content, emails, and browser history, disproportionately cutting workers on medical or parental leave (allegations Meta will contest in court, where the training data cannot follow it)
There was no breach. The keystrokes, the private conversations, the performance files — they simply landed someplace they shouldn't have landed, internally, where everyone could read them. A breach implies the data left the family.
Definitely Not the CTO
Chief Not-a-Breach Officer
Want Results Like These?
Schedule your complimentary Value Extraction Assessment today.