Claude Report — 2026-06-15
- Anthropic publishes official rebuttal to the Fable 5/Mythos 5 US government suspension, calling the demonstrated bypass "minor and previously known" — not grounds for a full commercial recall
- TCS × Anthropic: Claude deployed to 50,000 TCS employees across 56 countries in banking, insurance, healthcare, and public sector
- First Anthropic Public Record: only 15% of Americans trust AI companies — lowest of all institutions tested; 64% fear AI job displacement
- Agent SDK programmatic credit pool goes live today: Pro $20, Max 5x $100, Max 20x $200/month — separate from interactive usage, must be claimed, no rollover
- Claude Sonnet 4 and Opus 4 API endpoints retired today — all API requests using those pinned IDs now return errors
🚀 Anthropic Official
Statement on the US Government Directive to Suspend Fable 5 and Mythos 5
Published June 12; Anthropic confirms it received an export-control order at 5:21 PM ET suspending all user access to Fable 5 and Mythos 5 globally, and complied — while formally disputing the rationale. The company says the government's demonstration showed "minor, previously-known vulnerabilities" discoverable in other public models, argues that a "narrow potential jailbreak should not" trigger a full commercial recall, and warns the same standard would affect industry-wide deployments. All other Anthropic models remain available.
TCS × Anthropic — Claude Across 56 Countries in Regulated Industries
Published June 12; Tata Consultancy Services will deploy Claude to 50,000 employees across 56 countries and build industry-specific Claude solutions for financial services, healthcare, public sector, aviation, and telecom clients. Specific rollouts include claims-processing automation at Diligenta (22M+ policyholders), lending-advisory tools for banking, and Claude Code integrations for software engineering productivity; TCS iON will deliver Claude certifications across its 75M+ annual assessments platform.
Results from the First Anthropic Public Record
Published June 12; a ~52,000-person YouGov survey (fielded Nov–Dec 2025) finds a deep trust deficit: only 15% of Americans trust AI companies to make development decisions — the lowest rating of all institutions tested, trailing even government. Top fears are job displacement (64%), cognitive dependency (56%), and misinformation (52%); over 70% support government regulation, with priorities on privacy (56%), child safety (52%), and corporate liability (47%).
Agent SDK Programmatic Credit Pool — Now Live (June 15)
The billing change announced May 14 takes effect today: Agent SDK calls, claude -p, Claude Code GitHub Actions, and third-party Agent SDK apps (OpenClaw, Conductor, Zed, Jean) now draw from a separate monthly credit that mirrors each plan's price — Pro $20, Max 5x $100, Max 20x $200. Credits must be claimed via account email, reset monthly with no rollover, and are billed at API rates (Sonnet 4.6: $3/M input tokens, $15/M output); overflow routes to pay-as-you-go if extra usage is enabled, or halts until the next billing cycle if not.
Claude Sonnet 4 and Opus 4 API Retired (June 15)
claude-sonnet-4-20250514 and claude-opus-4-20250514 are now deprecated on the Anthropic API; all requests using these model IDs return an error as of today. Recommended migrations: Sonnet 4 → Sonnet 4.6; Opus 4 → Opus 4.6 (or Opus 4.7 for Adaptive Thinking). Claude.ai consumer accounts and Claude Code managed environments are not affected.
🔌 Claude Code Plugins
Plugin cache unavailable — top-10 tracking skipped this run.
🛠️ Skills
claude-code-system-prompts
Expands the documented prompt inventory from 350 to 515 strings (+165, the project's largest single update) for Claude Code v2.1.177 (June 12); new entries cover agent prompts, data templates, system reminders, and tool descriptions across all 27 built-in tools, reaching the most complete coverage of Claude Code's internal prompt surface to date.
Unchanged in window: anthropics/knowledge-work-plugins, glebis/claude-skills, anthropics/skills, ComposioHQ/awesome-claude-skills
🤖 Agents & Subagents
claude-agent-sdk-python v0.2.101
Fixes typed task_updated lifecycle events (June 13): exposes terminal system/task_updated events as a typed TaskUpdatedMessage object with task_id, patch, status, session_id, and uuid fields, plus a TERMINAL_TASK_STATUSES frozenset — consumers tracking active background tasks no longer hang indefinitely when a task finishes via a task_updated message without a corresponding TaskNotificationMessage.
Unchanged in window: claude-code-action (v1.0.148 maintenance, covered yesterday), nicobailon/pi-subagents, Chachamaru127/claude-code-harness
🔗 MCPs & Integrations
None of the top 5 had verifiable updates in the window.
Unchanged in window: modelcontextprotocol/servers, github/github-mcp-server (v1.3.0 shipped June 11 — one day outside window), modelcontextprotocol/typescript-sdk, awslabs/mcp, portainer/portainer-mcp
💡 Community — Workflows & Ideas
Two Claude Deadlines Today: Model Retirements + Billing Split
Enterprise DNA consolidates both June 15 milestones into a single actionable breakdown covering the exact model-string migrations (claude-sonnet-4-20250514 → claude-sonnet-4-6-20250718) and how to configure pay-as-you-go overflow for the Agent SDK credit pool to avoid halting automated workloads mid-cycle when credits deplete.
📰 Quick Mentions
Relevant links that don't warrant a section of their own:
- claude-agent-sdk-python v0.2.100 — June 12; bundles Claude CLI v2.1.176, no API-level changes
- Codersera: June 15 billing change walkthrough — step-by-step on claiming the new Agent SDK credit before it resets and verifying overflow settings
- ChatForest: Sonnet 4 & Opus 4 deprecation migration guide — one-liner model-string changes and common pitfalls for teams still on pinned API IDs