Valve's Steam Machine: Everything we know MetaAI is essentially the only AI assistant now available on WhatsApp. The EU could take interim measures against WhatsApp as it investigates AI providers' access to the app. On Monday, the EU's regulatory arm announced its preliminary view that Meta, WhatsApp's parent company, violated antitrust laws by blocking third-party AI assistants from operating on WhatsApp. The European Commission's is concerned that Meta's actions will limit competitors from entering the AI assistant market. We must protect effective competition in this vibrant field, which means we cannot allow dominant tech companies to illegally leverage their dominance to give themselves an unfair advantage, Teresa Ribera, executive vice-president for Clean, Just and Competitive Transition said in a statement. Ribera continued: AI markets are developing at rapid pace, so we also need to be swift in our action.

Instacart settles Federal Trade Commission's claim it deceived US shoppers Instacart has agreed to pay $60m in refunds to settle allegations brought by the United States Federal Trade Commission (FTC) that the online grocery delivery platform deceived consumers about its membership programme and free delivery offers. According to court documents filed in San Francisco on Thursday, Instacart's offer of "free delivery" for first orders was illusory because shoppers were charged other fees, the FTC alleged. "The FTC is focused on monitoring online delivery services to ensure that competitors are transparently competing on price and delivery terms," said Christopher Mufarrige, who leads the FTC's consumer protection work. An Instacart spokesperson said the company flatly denies any allegations of wrongdoing, but that the settlement allows the company to focus on shoppers and retailers. "We provide straightforward marketing, transparent pricing and fees, clear terms, easy cancellation, and generous refund policies -- all in full compliance with the law and exceeding industry norms," the spokesperson said.

Iteration 1 uses a broad, data-driven prior; subsequent iterations exploit memory to execute focused, theory-driven repairs, steadily converging on a causally defensible graph. This iterative loop is made explicit in Algorithm 1, while the statistics used during Evaluate are summarised in Table 2 and computed procedurally in Algorithm 2. 3.1. Causal Assumptions Every proposed DAG must explicitly address the four core assumptions required for causal identification. First, regarding unobserved confounding, the agent must state which latent factors remain and how observed variables serve as proxies for these unobserved influences. Second, the positivity assumption requires that the agent argue no sub-population is locked into or out of the treatment, often demonstrated by reporting overlap in the propensity-score distribution across treatment groups.

Recent advancements in Retrieval-Augmented Generation (RAG) have enabled Large Language Models to answer financial questions using external knowledge bases of U.S. SEC filings, earnings reports, and regulatory documents. However, existing work lacks systematic comparison of vector-based and non-vector RAG architectures for financial documents, and the empirical impact of advanced RAG techniques on retrieval accuracy, answer quality, latency, and cost remain unclear. We present the first systematic evaluation comparing vector-based agentic RAG using hybrid search and metadata filtering against hierarchical node-based systems that traverse document structure without embeddings. We evaluate two enhancement techniques applied to the vector-based architecture, i) cross-encoder reranking for retrieval precision, and ii) small-to-big chunk retrieval for context completeness. Across 1,200 SEC 10-K, 10-Q, and 8-K filings on a 150-question benchmark, we measure retrieval metrics (MRR, Recall@5), answer quality through LLM-as-a-judge pairwise comparisons, latency, and preprocessing costs. Vector-based agentic RAG achieves a 68% win rate over hierarchical node-based systems with comparable latency (5.2 compared to 5.98 seconds). Cross-encoder reranking achieves a 59% absolute improvement at optimal parameters (10, 5) for MRR@5. Small-to-big retrieval achieves a 65% win rate over baseline chunking with only 0.2 seconds additional latency. Our findings reveal that applying advanced RAG techniques to financial Q&A systems improves retrieval accuracy, answer quality, and has cost-performance tradeoffs to be considered in production.

Their novel solutions outperformed GPT -4, showcasing FinBen's potential to drive innovations in financial LLMs.

A key ingredient to the recent breakthroughs in language modeling has been the availability of large-scale datasets.

Loss Given Default (LGD) modeling faces a fundamental data quality constraint: 90% of available training data consists of proxy estimates based on pre-distress balance sheets rather than actual recovery outcomes from completed bankruptcy proceedings. We demonstrate that this mixture-contaminated training structure causes systematic failure of recursive partitioning methods, with Random Forest achieving negative r-squared (-0.664, worse than predicting the mean) on held-out test data. Information-theoretic approaches based on Shannon entropy and mutual information provide superior generalization, achieving r-squared of 0.191 and RMSE of 0.284 on 1,218 corporate bankruptcies (1980-2023). Analysis reveals that leverage-based features contain 1.510 bits of mutual information while size effects contribute only 0.086 bits, contradicting regulatory assumptions about scale-dependent recovery. These results establish practical guidance for financial institutions deploying LGD models under Basel III requirements when representative outcome data is unavailable at sufficient scale. The findings generalize to medical outcomes research, climate forecasting, and technology reliability-domains where extended observation periods create unavoidable mixture structure in training data.