We keep the receipts when companies silently make your stuff worse — pricing changes, rate-limit cuts, free-tier rugpulls, plan removals. Currently watching every major AI tool. Soon: gaming, streaming, and the rest.
On June 1, 2026 Copilot shifts from request-based to usage-based billing. Devs widely framed it as shrinkflation -- same price, less effective output.
Anthropic quietly removed Claude Code access from the $20 Pro plan, requiring Max ($100+/mo) for continued use. No proactive announcement.
Users discovered Anthropic no longer allows fixing API requests to a specific model snapshot, eliminating the primary defense against silent model swaps.
On April 2, 2026 OpenAI shifted Codex usage from a flat-tier model to token-based credits across most paid tiers, plus dropped Business from $25 to $20/seat the same day.
After weeks of community complaints about Claude Code ignoring CLAUDE.md guidelines, Anthropic acknowledged a quality regression in August 2025.
Anthropic announced weekly caps on Claude Code usage. The claim that fewer than 5% were affected was widely disputed; the HN thread hit 609 points and 705 comments.
Cursor's $20 Pro plan moved from 500 fast requests/month to a credit system that effectively cut usage to ~225 requests at the same price. CEO publicly apologized 3 weeks later.