Australia's Online Safety Amendment (Social Media Minimum Age) Act 2024 requires social media platforms to take 'reasonable steps' to prevent under-16s from having accounts, with enforcement being technology-neutral. This has raised concerns about the practical implementation and ambiguity around account deletion.

A research paper discusses the challenges of evolving skill modules in long-lived robot systems, including issues with versioning, rollback, and runtime governance. The paper presents negative results on the effectiveness of certain approaches and highlights the importance of proper termination conditions and acceptance criteria in achieving success.

This paper introduces Intent-Governed Access Control (IGAC), a server-side authorization layer for AI agents. IGAC generates a short-lived intent certificate from a trusted request, narrows the authorized tool manifest, and checks proposed tool and payload effects before execution. This approach prevents static permissions from exceeding user requests and relies on certificate fidelity and sound effect bounds.

A new benchmarking framework is proposed for evaluating the explanatory quality of open-weight vision-language models in face recognition. The framework focuses on explanation quality, including relevance and faithfulness, and provides a structured explanation format for automated querying and auditing. The authors benchmark several families of open-weight VLMs and highlight the need for explanation quality metrics in face recognition systems.

Researchers proposed a method to link source code to scholarly articles in Wikidata, creating a knowledge graph of software heritage. The method harvests and validates publication-to-repository pairs from specific sources, yielding a corpus of 4,397 pairs. The approach enables rule-based reconciliation of software and articles, with a read-only lookup against Wikidata showing 4,182 new software items created.

Researchers have developed KG-Chronos-2, a knowledge-graph-augmented model for surrogate forecasting in HEC-RAS water-surface elevation tasks. The model combines a frozen Chronos-2 predictor with graph-conditioned historical retrieval and input-aligned correction, achieving improved accuracy compared to other models. This development may be useful for people building or operating AI agents that involve water-surface elevation forecasting or similar tasks.

EnterpriseVal is a use-case-level evaluation system for generative AI (GenAI) in the enterprise. It addresses the issue of measuring GenAI efficacy, reliability, and value by providing a formal specification, metric catalogue, grading protocol, and decision-making framework. The system aims to help organizations make informed decisions about GenAI deployment and scaling. A pilot across three workflows in a global bank showed promising results, including high citation precision and reduced analyst refinement effort.

MIST is a multimodal survival prediction model that combines genomic and histology data for cancer outcome prediction. It uses a compact and effective strategy called genomic-guided histology attention to enrich molecular information with histology context.

A new risk-aware occupancy method, risk-aware occupancy, is proposed for end-to-end autonomous driving. It encodes global scene occupancy, map-derived traffic constraints, and future dynamic agent occupancy into a unified BEV map, capturing risk evidence for trajectory planning. A neural network, ROIDrive, predicts risk-aware occupancy and injects it into planning queries for safety-oriented trajectory generation.

LEGIT is a credentialing protocol for trustworthy AI agent marketplaces, connecting certification, reputation, and proposed marketplace allocation. It binds measured quality and cost per task to an agent configuration, task domain, evaluation budget, and evidence, and provides a way to verify credential records and inspect optional visual profiles.

A new framework, PlaceReasoner-Beta, tackles macro placement in VLSI physical design by reformulating it as a closed-loop reasoning problem. It uses a multi-agent approach with a vision-language model, geometric verifier, physical verifier, and post-route optimizer to improve layout quality and timing. A benchmark, PlaceReasoner-Bench, is introduced for reproducible evaluation.

Cloudflare has announced the general availability of Python Workers, allowing developers to run Python applications in the Cloudflare Workers runtime. This makes Python a first-class language on the Cloudflare Developer Platform, enabling seamless integration with other Cloudflare services and popular Python frameworks like FastAPI, Django, and Flask.

A new workflow, VLM-CAD, combines vision language models with a neuro-symbolic structural parsing module to improve analog circuit design by leveraging multimodal reasoning and providing explainability. This workflow, designed for engineers, addresses spatial blindness and logical hallucinations in VLMs when interpreting complex engineering content.

Researchers propose a federated learning framework for video moderation called FedVideoMAE, which uses differential privacy and secure aggregation to protect user data. This framework is designed for edge and privacy-sensitive settings, reducing the need for raw video data to be sent to a central server.

Researchers propose a new method, Sampling-Guided Policy Search (SGPS), to accelerate visual policy learning for AI agents. SGPS combines sampling-based model-predictive control with first-order policy optimization to learn policies for locomotion and manipulation. This approach is shown to improve policy learning and enable zero-shot transfer to real-world robots.

This paper proposes a scalable trust discovery architecture for the Internet of Agents, addressing the need for agent registration, trustworthy identification, and capability-oriented discovery. The proposed architecture consists of three layers: Agent Root, Agent Registry, and Agent Resolver, and incorporates a registry-suffix-anchored composite identity scheme and dual-certificate authentication. A prototype implementation demonstrates the feasibility of the architecture, achieving low latency and high registration and discovery rates.

A new platform-mediated mechanism, PAPC, is proposed to manage privacy-propagation externalities in AI-mediated workflows. PAPC intercepts information-moving events to prevent downstream exposure costs and preserves deterministic task completion. This development is relevant to people building and operating AI agents, as it addresses a key challenge in ensuring the privacy of AI-mediated systems.

A study evaluates the effectiveness of individual components in coding harnesses for AI agents. The researchers found that context management becomes more valuable as the context-window budget tightens, and that staging rule-based elision before LLM-based summarization provides the strongest efficiency among context-management strategies. The study also found that planning shifts from an accuracy scaffold for weaker models to a cost saver for stronger models. These findings inform model- and budget-aware harness design and provide a modular framework for evaluating future harness components.

Researchers present NeuSOGA3D, a hybrid framework that combines neural implicit methods with explicit symbolic geometric reasoning for 3D geometric reconstruction from unorganized point clouds. This approach provides interpretable and reusable geometric representations, which can be beneficial for AI agents working with 3D data.

A new model, PaGNet, is proposed for forecasting corporate tax avoidance proxies from firm-year panel data. It combines a LightGBM branch with a Panel-MLP branch, using attention-pooled temporal aggregation and shared-trunk multi-task learning. The model is evaluated on a dataset of Korean listed firms, showing improved performance over six baselines. The model's branch-reliance diagnostic provides additional insights into the forecasting process.