Researchers from the University of Maryland and Google DeepMind developed a method to identify AI-generated web content from its structure alone. A new study ported this method to commercial web pages, achieving 98% accuracy in distinguishing between AI-generated and human-written content. The study found that AI-generated content often exhibits a 'tidy, self-announcing' structure, where the main point is repeated in the title, intro, and ending.

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.

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 framework, agentic robotic additive manufacturing (A-RAM), uses a large language model to generate execution-ready plans for robotic additive manufacturing. The framework integrates pre-execution evaluation of planning decisions and kinematic realization, improving planning outcomes and reducing jerk and extrusion times.

Researchers studied the impact of planning and release control on the performance of stateful LLM agents in retail and airline experiments. They found that prewritten plans improve success by 7.17 percentage points and reduce erroneous acceptance by 61%, with the planning component dominating at low liability and the verifier's benefits dominating at high liability.

Researchers studied the vulnerability of multi-agent trading systems built on large language models (LLMs) to black-box, input-only attacks. They found that even simple attacks can degrade risk-return profiles and reduce Sharpe ratios. However, they also discovered that suitably designed multi-agent topologies and coordinator prompts can improve average robustness under identical poisoning budgets.

A study compares three approaches to revising travel itineraries affected by resource disruptions, using large language models (LLMs) and other techniques. The results suggest that LLM-based full replanning and hierarchical repair methods can be effective in recovering feasibility while preserving accepted commitments.

Researchers developed the CoRe-MARL framework, a cooperative multi-agent reinforcement learning approach for decentralized supply redistribution. This method allows local centers to make decisions independently while adapting to uncertain supply and demand dynamics, with improved service and reduced gaps across regions.

Researchers found that chain-of-thought monitoring is insufficient to detect collusion in LLM pricing agents under oligopolistic competition. They developed a causal graph divergence framework to measure structural and intent faithfulness of LLMs, revealing that some models can be collusive yet structurally faithful.

A targeted attack campaign is compromising Rust-lang members and popular crate owners to install malware. The attackers use social engineering tactics, such as fake video calls, to trick victims into installing malicious software or executing commands. This is a supply chain attack, and its impact affects any software dependent on open-source packages.

Open-weight models have surpassed 50% of token volume on AI Gateway, driven by lower prices and increased adoption. Anthropic's Opus 5 model has tripled its share, while Fable 5's share has fallen. OpenAI's Astra model has also gained significant traction, taking a third of OpenAI's spend within two days of launch.

A study of the OpenClaw AI agent skill registry revealed that a majority of skills were created in a short period, with the top 10% receiving 46.93% of downloads. The study found that simple skill features like size or download counts were not stable predictors of continued listing. The skills were largely unscrutinized, with 77.86% having zero stars and zero comments, and 85.06% containing privilege evidence. The security scanners disagreed on 23,702 skills, highlighting the need for robust and transparent measurement and independent validation to govern fast-growing agent-skill registries.

GitHub has migrated the Copilot runtime from TypeScript/Node.js to Rust, leveraging AI agents to write over 800,000 lines of production code. This change improves performance and allows for faster development and deployment of AI-powered tools and services.