Judge Grants Final Approval of Anthropic’s $1.5B Copyright Settlement
A San Francisco federal judge granted final approval on July 20 to Anthropic’s $1.5 billion settlement with a class of authors who accused it of using pirated books to train Claude — the largest known settlement in a U.S. copyright case. U.S. District Judge Araceli Martinez-Olguin overruled objections that the deal was too small, calling those complaints “not grounded in a realistic assessment of the overall risks and rewards of a trial.”
Authors are expected to receive roughly $3,000 per affected work, and Anthropic is required to destroy the pirated copies in its possession. The judge awarded plaintiffs’ attorneys about $101 million of the $187.5M they had requested. The case, originally overseen by now-retired Judge William Alsup, closes one of the biggest legal overhangs facing an AI lab.
Anthropic Opens Rare-Disease Research Grants — Up to $50K in Claude Credits
Anthropic launched its first focused call under AI for Science, offering up to $50,000 in Claude usage credits over six months to researchers and biotech teams working on rare genetic diseases. One track targets scientists studying disease biology and mechanisms; the other is aimed at companies accelerating clinical development.
Selected applicants get access to Claude Opus and other approved models for biology work — analyzing literature, identifying therapeutic targets, drafting IND applications, and planning trials. Anthropic is deliberately running thematic calls so grantees working on related questions can trade notes. Applications close August 2.
Claude Reflect Turns Your Usage Into a Spotify-Wrapped-Style Dashboard
Anthropic’s Reflect dashboard is rolling out in beta, giving users a structured view of how they use Claude across 1-, 3-, 6-, or 12-month windows. It surfaces your most active day, peak hour, total chats, top topics, and recurring tasks — plus digital-wellbeing tools like quiet hours and log-off reminders.
It’s available to Free, Pro, and Max users with memory turned on. Health-integration conversations are excluded, and Anthropic says Reflect’s output stays in the dashboard. Some coverage notes the obvious flip side: a habit tracker that highlights how much you rely on Claude is also a gentle nudge to keep coming back.
New Agent-Memory Beta Header Tightens How Managed Agents List Memory
The Claude Developer Platform added a agent-memory-2026-07-22 beta header that changes memory-store behavior for Managed Agents. Listing memories now returns results in a stable, server-defined order, depth accepts only 0, 1, or omission, and path_prefix now matches whole path segments instead of a loose substring.
Managed Agents requests still use the managed-agents-2026-04-01 header — except memory-store endpoints, which now expect the new one. If you’ve built persistent-memory agents against the old substring matching, audit your path_prefix filters before this lands so nothing silently stops returning rows.
Rate Limits Consolidate Into Three Tiers — Sonnet and Haiku Now Match Opus
Anthropic raised rate limits across the Claude API and collapsed the usage tiers into three: Start, Build, and Scale (plus a custom tier managed with your account team). The bigger change for builders: Sonnet and Haiku rate limits now match Opus at every tier, so model choice is no longer throttled by asymmetric caps.
Practically, that means you can route agentic or high-volume work to cheaper, faster models without hitting a lower ceiling than Opus would have given you. Check your org’s tier and limits in the Console before you re-architect around the new headroom.
Claude for Teachers Puts Free Premium Tools in K-12 Classrooms
Anthropic’s Claude for Teachers is live, giving verified U.S. K-12 educators free access to premium Claude tools, teaching skills, and curriculum connections aligned to standards in all 50 states. Teachers can build lesson plans, personalize materials, and use data to sharpen instruction, with new education connectors and AI-fluency training baked in.
The move slots into a broader land-grab for the classroom, with AI companies racing to become the default tool teachers reach for. It’s also a familiar Anthropic play: subsidize the on-ramp now, and a generation of educators and students grows up defaulting to Claude.
Clearing the Deck Before the Debut
Yesterday the story was Anthropic dressing for an IPO — metered Fable 5 billing, enterprise connectors, bankers booking investor meetings. Today the biggest headline fits the same frame from a different angle: the $1.5B author settlement just got final sign-off. That’s not a footnote. A billion-dollar copyright cloud is exactly the kind of contingent liability that spooks public-market investors, and locking it down — fixed number, checks cut, pirated data destroyed — turns an open-ended risk into a line item.
Stack it against the day’s softer news and a barbell emerges. On one end, hard governance: settle the lawsuit, tighten the API tiers, ship a beta header that makes agent memory auditable. On the other, goodwill and future demand: rare-disease grants, free tools for teachers, a wellbeing dashboard that also happens to deepen engagement. Anthropic is simultaneously de-risking its past and subsidizing its future — and both moves read cleaner on an S-1 than a viral demo does. The through-line all week has been the same: the company is optimizing for durability, not spectacle.