The Enshittification of Generative AI

Source: Where’s Your Ed At Date: 2025-08-08 URL: https://www.wheresyoured.at/the-enshittification-of-generative-ai/

Summary

Ed documents AI product enshittification in progress: OpenAI has removed model choice from paid tiers, restricted reasoning capabilities to 200 messages weekly for Plus subscribers (down from hundreds daily), sunset older models, and doubled API pricing while adding “priority processing” fees. The pattern is textbook — initial generous offerings built user dependency, now extraction begins. With Anthropic burning cash on Claude Code at $33M monthly before implementing rate limits, Ed’s claim is that all the major AI labs are degrading their products in parallel as unsustainable burn rates force the pivot to tiered extraction.

Implications

• Enshittification. This is Doctorow’s framework applied in real time: AI companies first made products good to attract users and developers, now they’re making them worse to extract maximum revenue. The rate limit and tier restriction data are the receipts.
• AI financial sustainability. The enshittification move is evidence that the underlying economics never worked — if they did, there’d be no need to degrade the product. $33M monthly Claude Code revenue requiring rate limits means the cost structure is still broken at scale.
• Generative AI ROI. Enterprises that built workflows around specific model capabilities and pricing assumptions now face moving goalposts. The business case built on API pricing from 2023 doesn’t survive 2x price increases plus capability restrictions.