Lawful AI Daily Brief — 2026-04-16

🛠️ Tool Updates

• Claude Code v2.1.110 dropped a proper operator starter pack: /tui fullscreen, /focus, mobile push notifications, cleaner /plugin triage, and scheduled tasks that wake back up on --resume.
• Hooks/MCP reliability got a polish pass: fewer permission-hook weirdos, fewer disconnect hangs, less “why is this frozen?” energy.
• Codex rust-v0.121.0 added codex marketplace add (GitHub/local/URL), tighter MCP namespacing, improved Ctrl+R history in TUI, plus stronger sandbox/devcontainer hardening.

💡 Tip of the Day

If your legal/research prompts keep losing context, force a longer prompt-cache TTL and regain your sanity:

# keep context warm longer for heavy legal workflows
export ENABLE_PROMPT_CACHING_1H=1

claude
/tui fullscreen
/focus

Bonus move for MCP-based compliance stacks:

servers:
  legal-regs:
    command: uvx
    args: [eurlex-mcp]
    metadata:
      jurisdiction: EU
      retention_days: 30
      lawful_basis: "Art.6(1)(f)"

⚖️ Legal x AI Watch

• GitHub activity around EU AI Act / AI regulation topics stayed lively, with newly pushed repos focused on:
  • open governance knowledge packs,
  • compliance guardrails for agent systems,
  • enterprise governance toolkits mapping to standards like ISO/IEC 42001 and AI risk frameworks.
• Practical takeaway: teams are investing in compliance-by-design scaffolding (controls, documentation, governance templates) rather than one-off policy docs.

📚 Fresh Papers

• From Anchors to Supervision: Memory-Graph Guided Corpus-Free Unlearning for LLMs — proposes corpus-free unlearning for removing memorized sensitive/copyrighted content (very compliance-relevant).
  http://arxiv.org/abs/2604.13777v1
• RPS: Information Elicitation with Reinforcement Prompt Selection — reinforcement-based prompting to elicit latent user info in dialogue workflows.
  http://arxiv.org/abs/2604.13817v1
• From P(y|x) to P(y): RL in Pre-train Space — explores pushing optimization into pre-train space to expand reasoning performance.
  http://arxiv.org/abs/2604.14142v1
• From Feelings to Metrics: Formalizing Vibe-Testing — maps subjective “vibe checks” to measurable LLM evaluation signals.
  http://arxiv.org/abs/2604.14137v1

🔥 Trending Repos

• thegdsks/awesome-modern-cli (⭐ 31) — curated modern CLI alternatives.
• Mondo-Robotics/DiT4DiT (⭐ 14) — vision-action model framework.
• zhengnaichuan2022/PAS-Net (⭐ 13) — physics-aware spiking NN for HAR.
• ChaosRealmsAI/NextFrame (⭐ 11) — Rust AI video timeline engine.

AI/LLM heavyweight pulse (still dominating): AutoGPT, dify, langchain, firecrawl, ragflow, mem0.

🧍 Standup One-Liner

We upgraded the AI engineering stack for fewer ops headaches, spotted fresh compliance-first repo momentum, and flagged new unlearning research that actually matters for legal risk.

---

🔗 Repo: https://github.com/laugustyniak/lawful-ai-staging