Lawful AI Daily Brief — 2026-04-19

🛠️ Tool Updates

• Claude Code (v2.1.113 / v2.1.114) shipped stability + security polish: permission-dialog crash fix, safer Bash matching, better /loop wakeup control, and an MCP concurrent-call timeout fix.
• Codex CLI alpha train (0.122.0-alpha.8/.9/.10) pushed practical upgrades: better MCP stdio wiring, plugin-cache crash hardening, high-detail image defaults, and model metadata with max context windows.
• Ecosystem pulse: awesome-claude-code activity looked mostly like data refresh churn (quiet on brand-new tools).

💡 Tip of the Day

If your agent stack touches sensitive data, treat network egress policy like a seatbelt, not a “nice to have.”

[sandbox.network]
deniedDomains=["*.pastebin.com","*.temp-mail.org","*.anonfiles.com"]

Bonus move: include a quick “who called what” tool log in CI artifacts so compliance evidence is not a panic project later.

⚖️ Legal x AI Watch

• Compliance-flavored repos with fresh activity in the last day:
  • SDL-HQ/sir-firewall — deterministic pre-inference governance gate + signed audit trail framing.
  • atharvajoshi01/finreg-ml — regulation-aware finance ML pipeline with explainability/fairness hooks.
  • ZOLAtheCodeX/aigovops — operations catalog for governance frameworks (EU AI Act, ISO/IEC 42001, NIST AI RMF).
  • sanna-ai/sanna — trust/compliance infrastructure for agents with cryptographic receipts.

📚 Fresh Papers

• GDPR Auto-Formalization with AI Agents and Human Verification — proposes automated GDPR formalization using LLMs + human verification loop.
  https://arxiv.org/abs/2604.14607v1
• BenGER: A Collaborative Web Platform for End-to-End Benchmarking of German Legal Tasks — legal-task benchmarking platform for LLM reasoning evaluation.
  https://arxiv.org/abs/2604.13583v1
• From Anchors to Supervision: Memory-Graph Guided Corpus-Free Unlearning for Large Language Models — corpus-free unlearning method relevant to privacy/copyright-sensitive memorization.
  https://arxiv.org/abs/2604.13777v1

🔥 Trending Repos

1. Significant-Gravitas/AutoGPT
2. f/prompts.chat
3. langgenius/dify
4. langchain-ai/langchain
5. firecrawl/firecrawl

🧍 Standup One-Liner

Today’s vibe: agent tooling got sturdier, legal-tech repos stayed busy, and compliance moved one inch closer to “boring by design.”

---

Source & archive: https://github.com/laugustyniak/lawful-ai-staging