🌅 AI Daily Digest — April 01, 2026

Data Pulse

Copaw-9B (Qwen3.5 9b, alibaba official agentic finetune) — Alibaba has released Copaw-9B, an agentic fine-tune of the Qwen3.5-9B large language model. The update focuses on enhancing autonomous decision-making capabilities for enterprise applications.
Mercor says it was hit by cyberattack tied to compromise of open-source LiteLLM project — Mercor.io Corporation, an AI hiring startup, disclosed a cybersecurity incident affecting its operations. The breach is linked to a compromised open-source LiteLLM project, impacting its founders’ billionaire status.
Newsom signs executive order requiring AI companies to have safety, privacy guardrails — California Governor Gavin Newsom signed an executive order mandating AI companies to implement safety and privacy guardrails. The requirement applies to all AI systems operating within the state.
The Claude Code Source Leak: fake tools, frustration regexes, undercover mode — Anthropic PBC faced a setback after the public leak of the source code for its Claude Code CLI application. The incident exposed vulnerabilities in its development tools and workflows.
The Download: AI health tools and the Pentagon’s Anthropic culture war — Anthropic faced legal and PR turmoil as a California judge issued a preliminary injunction blocking the Pentagon’s attempt to label it a supply chain risk. The case highlights tensions over AI governance and national security.

Reasoning-Driven Synthetic Data Generation and Evaluation — Tim R. Davidson, Benoit Seguin, Enrico Bacis. Although many AI applications of interest require specialized multi-modal models, relevant data to train such models is inherently scarce or inaccessible.
From Skeletons to Semantics: Design and Deployment of a Hybrid Edge-Based Action — Ganen Sethupathy, Lalit Dumka, Jan Schagen. Public spaces such as transport hubs, city centres, and event venues require timely and reliable detection of potentially violent behaviour to support public safety.
Tracking vs. Deciding: The Dual-Capability Bottleneck in Searchless Chess Transf — Quanhao Li, Wei Jiang. A human-like chess engine should mimic the style, errors, and consistency of a strong human player rather than maximize playing strength.
TSHA: A Benchmark for Visual Language Models in Trustworthy Safety Hazard Assess — Qiucheng Yu, Ruijie Xu, Mingang Chen. Recent advances in vision-language models (VLMs) have accelerated their application to indoor safety hazards assessment.
CausalPulse: An Industrial-Grade Neurosymbolic Multi-Agent Copilot for Causal Di — Chathurangi Shyalika, Utkarshani Jaimini, Cory Henson. Modern manufacturing environments demand real-time, trustworthy, and interpretable root-cause insights to sustain productivity and quality.

How to Benchmark AI Models with MLPerf 2.0 This tutorial explains the significance of standardized benchmarking practices and highlights common flaws in current AI evaluation methods. Readers will gain insights into improving industry reliability through MLPerf 2.0’s structured approach.
How to Implement a Custom Claude API Wrapper with Python The guide details a technical error encountered during API integration, emphasizing lessons learned from the process. It provides a practical example of building a wrapper while acknowledging its limited industry impact.

Head of Finance North America at Adyen (San Francisco)
Senior Manager Strategic Partner Marketing at Vanta (Remote)
Senior Staff Software Engineer AI Customer Operati at Monzo (Cardiff, London)

New this week: