OpenAI’s AI program backs independent journalism in Ukraine
OpenAI’s program with AIRPPU and WAN-IFRA is aimed at helping Ukrainian news organizations strengthen innovation, resilience, and independent reporting workflows. The initiative focuses on practical capacity building so editorial teams can use AI more safely while maintaining sovereignty over how they produce and validate content.
Corporate Language Model (CLM) turns enterprise knowledge into sovereign, auditable action
arXiv’s CLM framework argues that enterprise AI failures usually stem from missing decision and execution “substrates,” not from the base model being too weak. It proposes an architecture that grounds firm-specific multimodal and tacit knowledge in an ontology, then uses a security layer plus spec-as-code to make reasoning output traceable and executable under human oversight.
Seattle Times and Newsday sue OpenAI and Microsoft for alleged copyright infringement
Seattle Times and Newsday are seeking legal remedies alleging OpenAI and Microsoft used their journalism as training data without permission and that the systems can reproduce passages in response to user queries. The complaint adds Microsoft because Copilot relies on OpenAI technology, positioning the case as part of a wider wave of media lawsuits.
From Matching Models to Recruiting Agents: AI recruitment workflows and what still breaks
This systematized narrative review tracks how recruitment automation evolved from ranked profile matching into multi-stage, tool-using agent workflows that retrieve evidence and support contested decisions. It also highlights why evaluation keeps falling short – behavioral labels can be confounded, synthetic/private data limits real-world validity, and most benchmarks fail to jointly measure utility, fairness, privacy, and security.
From Matching Models to Recruiting Agents: AI recruitment workflows and what still breaks (evaluation evidence map)
The paper’s key claim is that evaluation needs to be anchored to “what evidence the system actually uses” across the pipeline, not just to final ranking accuracy. It proposes a staged mapping from evaluation evidence to the strongest defensible claims, pushing recruitment systems toward reciprocal, temporally controlled, evidence-grounded governance.