2026 · 05 · 07 · Thu

Updates for 5/7

Anthropic doubled Claude Code rate limits on all paid plans after securing 220,000+ GPUs. A 2025 case adds a new risk: discipline for unverified AI output.

A · Theme of the day

Claude Code doubles rate limits, keeps prices flat

After securing SpaceX's Colossus-1, Claude Code limits doubled on paid plans.

Claude Code's rate limits double, peak slowdowns gone

Free vs Pro vs Team vs Enterprise — A Thorough Plan Comparison

Compared to before

For months, daily Claude Code users on Pro hit slowdowns after 5 PM and throttling late in the month. Compute was rationed at peak, so heavy users got hit first.

What changed

[May 2026 Update] After Anthropic secured SpaceX's Colossus-1 (220,000+ GPUs), Claude Code rate limits doubled on all paid plans and peak-hour limits ended.

Why it matters

If you run Claude Code most workdays, this is a free capacity upgrade—no more rationing prompts. If you open Claude once or twice a week, the change is invisible.

Source: localhost　·　→ View in AI Encyclopedia

B · Theme of the day

Lawyers face disciplinary risk when AI output goes unchecked

A 2025 case of AI hallucinations in a court brief creates new disciplinary risk.

Lawyers face disciplinary risk when AI hallucinates in court filings

Generative AI Litigation Tracker 2026: NYT vs OpenAI, Getty vs Stability, RIAA vs Suno

Compared to before

Until recently the dominant AI-legal story was copyright (NYT v. OpenAI, Getty v. Stability). A hallucinated citation meant embarrassment, not formal discipline.

What changed

In 2025, Latham & Watkins filed a brief in Concord Music Group v. Anthropic containing AI hallucinations, raising disciplinary risk for unverified AI output.

Why it matters

For lawyers and anyone filing professional documents, skipping verification now risks bar complaints or sanctions. If AI only drafts internal notes, exposure stays low.

Source: localhost　·　→ View in AI Encyclopedia

Past updates

A daily archive of changes actually applied to the site.