2026-07-21 – Teachers Data Policy Clarified & Tighter Subagent Controls

🧭 Claude for Teachers Clarifies: Data Policy Is Set by Your District, Not Anthropic

A week after launch, Anthropic updated the Claude for Teachers page with a clarifying note: whether and how educational data may be used within the product is determined by district and state policies, not a single uniform Anthropic policy. For a free K-12 tool operating across all 50 states' varying education-data regulations, this is a necessary clarification rather than a walk-back — US student data privacy law (FERPA and state-level equivalents) genuinely does vary by jurisdiction, and a one-size-fits-all data policy would have been inaccurate.

Practical implication for educators evaluating the tool

Teachers considering Claude for Teachers should check their own district's data policy for AI tools before assuming Anthropic's terms are the final word — the actual constraints on what student data can touch the tool are set locally, and district IT/compliance teams are the authoritative source, not the product's general terms page.

⭐⭐⭐ anthropic.com

🧭 Claude Code Adds Emoji Autocomplete, Clearer Transcript Warnings & Tighter Subagent Controls

A Claude Code update brings emoji shortcode autocomplete, clearer warnings when a transcript is about to be written to disk, and tighter controls for subagents, budgets, and background sessions — giving teams finer-grained limits on how much a spawned subagent tree can consume before requiring explicit approval to continue.

Worth reviewing your subagent budget defaults

Teams running nested sub-agent workflows (a pattern that's grown considerably since it shipped in June) should check whether the new, tighter default controls change behaviour for existing automations — a budget cap that was previously implicit may now be explicit and enforced differently.

⭐⭐⭐ code.claude.com