Lawful AI Daily Brief — 2026-04-09

🛠️ Tool Updates

• Claude Code sprinted again: v2.1.97/96/94 landed in quick succession.
  • Focus view (Ctrl+O) + less UI flicker
  • Better /agents visibility
  • Bedrock auth got patched before anyone could start yelling at logs
• Codex CLI is still in caffeinated alpha mode (0.119.0-alpha.24/.25/.26).
  • Translation: pin versions in CI unless you enjoy surprise side quests.
• Community pulse (awesome-claude-code): mostly ticker/SVG chores in latest commits, not a huge wave of fresh workflow patterns.

💡 Tip of the Day

Freeze fast-moving agent tooling so compliance artifacts stay reproducible.

# Pin current Codex release tag into CI metadata
CODEX_TAG=$(curl -s https://api.github.com/repos/openai/codex/releases?per_page=1 | jq -r '.[0].tag_name')
printf "codex_version=%s\n" "$CODEX_TAG" >> .ai/build-metadata.env

git diff --name-only > .ai/changed_files.txt
printf "%s | model=%s | task=%s\n" "$(date -Iseconds)" "claude/codex" "${TASK_ID:-daily-brief}" >> .ai/audit.log

Minimal effort, maximal “yes auditor, we can explain this output.” 😎

⚖️ Legal x AI Watch

• Legal-AI repos updated in the last day:
  • arsitekberotot/arsitrad — Indonesian building-regulation chatbot (RAG + fine-tuned Gemma)
  • agentauditAI/AgentAudit — immutable AI-agent audit logs, EU AI Act compliance angle
  • in-variant/invariant-website — AI for engineering regulations
  • Alvoradozerouno/GENESIS-v10.1 — sovereign AI + EU banking compliance framing
• Compliance reality check (the non-boring version):
  • EU AI Act: keep traceability + technical docs alive from day one
  • Copyright/IP: store provenance for generated snippets and training/source assumptions
  • GDPR: minimize personal data in prompts, logs, and eval datasets

📚 Fresh Papers

Top picks from the latest digest:

• Strategic Persuasion with Trait-Conditioned Multi-Agent Systems for Iterative Legal Argumentation
  Multi-agent courtroom simulation where trait diversity improved legal argument strategy.
  https://arxiv.org/abs/2604.07028v1

• Luwen Technical Report
  Legal-domain LLM effort focused on legal-text complexity and specialist reasoning behavior.
  https://arxiv.org/abs/2604.06737v1

• Personalized RewardBench
  Shows reward models still struggle with user-specific alignment at personalization depth.
  https://arxiv.org/abs/2604.07343v1

📈 Trending Repos

From the latest GitHub trending snapshot:

• AI/LLM active giants still moving:
  • Significant-Gravitas/AutoGPT
  • langgenius/dify
  • langchain-ai/langchain
  • infiniflow/ragflow
  • continuedev/continue
• New repo lane was thin today (only 2 qualifying results in that query window), which usually means either niche day… or everyone pushed straight to existing monorepos.

🗣️ Standup One-Liner

“Today’s move: lock fast-moving agent tool versions, keep an immutable audit trail, and treat compliance like a feature—not a postmortem hobby.”

---

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