New research paper introduces CAGE-1 evaluation framework for auditing autonomous AI agents in enterprise workflows, addressing control, authorization, policy compliance, and accountability challen...
Eticas proposes open infrastructure for operationalizing AI audits across 74 competing risk taxonomies, moving beyond naming risks to executing measurable, defensible tests against real systems.
Researcher proposes meta-benchmarking framework organizing 452 public LLM benchmarks into 41 work activities and 38 banking domains, showing that general-purpose leaderboards miss domain-specific d...
Researchers propose active learning strategy for cascaded table detection and structure recognition in business documents, reducing costly fine-grained annotation burden while maintaining pipeline ...
SpreadsheetBench 2 evaluates AI agents on end-to-end spreadsheet tasks (generation, debugging, visualization) used in business financial modeling and reporting—moving beyond isolated cell operation...
Academic research reveals that LLM-compressed financial data loses decision-critical fidelity, with errors potentially amplifying across agentic systems—a risk for automated investment analysis.
Researchers propose diffusion-guided learning to improve fraud detection in transaction networks despite sparse labels and imbalanced data, addressing real-world challenges in financial system secu...
Researchers propose probabilistic multi-agent systems to automate labor-intensive IT-Grundschutz audits under NIS-2 Directive, reducing manual documentation and validation burden for compliance.
Researcher evaluates Claude and ChatGPT on IPO financial analysis using S-1 filings, finding Finance Agent v2 inadequate for forward-looking IPO tasks; proposes automated rubric generation as impro...
Research paper evaluates frontier LLMs (Claude, ChatGPT) on IPO financial analysis tasks using automated rubric generation, extending Finance Agent v2 benchmarks to SEC S-1 filings and SpaceX IPO c...
Academic paper argues AI agents need intent-governed authorization beyond static credentials to prevent unauthorized tool calls in accounting workflows and multi-app integrations.
VADAOrchestra combines LLM-based agentic reasoning with symbolic logic to enable adaptive, auditable decision-making in business processes—addressing a core tension in AI-driven accounting workflows.
VADAOrchestra combines symbolic logic with LLM-based reasoning to enable adaptive business workflows that maintain auditability—addressing a key gap for regulated industries like accounting and fin...
Meridian releases Spreadsheet Arena, an open benchmarking platform co-developed with Cornell, CMU, and Scale AI to evaluate LLM performance in generating spreadsheet workbooks through blind pairwis...
An audit of 1 billion tokens reveals a major gap between token compression rates (99%) and actual billing, suggesting potential overcharges or token counting inconsistencies in LLM API pricing.
Blue J and CPA.com survey shows tax firms have moved from experimenting with AI to embedding it as core workflow—adoption rate nearly doubled year-over-year.
AgentFinVQA combines on-premise deployability with auditability for financial chart QA, addressing regulated auditors' need for explainable AI without accuracy compromise.
Stanford researchers release SEFD, a layout-faithful, open-source dataset of 4.4M SEC filings reconstructed into MultiMarkdown format for pretraining financial LLMs at scale.
Stanford researchers publish SEFD, an open-source dataset reconstructing SEC EDGAR filings into layout-faithful MultiMarkdown to address scarcity of high-quality long-context training data for fina...
Research paper proposes multi-vector AI agent for banks to detect both signature-based fraud (card attacks, account takeover) and behavioral financial crime (structuring, BEC, money laundering) sim...