Global AI Weekly

Issue number: 130 | Tuesday, December 30, 2025

Highlights

Enterprise AI Showdown: Anthropic vs OpenAI

Enterprise AI Showdown: Anthropic vs OpenAI

Anthropic has introduced open-source Agent Skills, aiming to provide enterprise AI tools as a competitive alternative to OpenAI's offerings. These tools are designed to allow businesses to create and customize AI systems that align with their specific needs and values. By making these tools open-source, Anthropic is encouraging innovation and broader access in the AI landscape.

techradar.com

ChatGPT launches a year-end review like Spotify Wrapped

ChatGPT launches a year-end review like Spotify Wrapped

ChatGPT has introduced a fun year-end review feature, similar to Spotify Wrapped, where users can revisit their most memorable moments with the AI. The review includes highlights such as awards, poems, and even pictures that reflect the user's year of interaction with ChatGPT. It's a creative way for users to look back on their unique chats and experiences over the past year.

techcrunch.com

Critical n8n Flaw (CVSS 9.9) Enables Arbitrary Code Execution Across Thousands of Instances

Critical n8n Flaw (CVSS 9.9) Enables Arbitrary Code Execution Across Thousands of Instances

A critical vulnerability (CVE-2025-68613) has been discovered in n8n, allowing authenticated users to execute arbitrary code. The issue affects versions 0.211.0 to 1.120.4, posing a significant risk with a CVSS score of 9.9. Users are urged to upgrade to the latest patched version to secure their systems.

thehackernews.com

Research

LLMs work better together in smart contract audits

LLMs work better together in smart contract audits

Academic research highlights that combining fine-tuned language models with ensemble voting significantly enhances the detection of vulnerabilities in smart contracts. This approach leverages the strengths of multiple models working together, resulting in more accurate and reliable audits. The study emphasizes the potential of this collaborative method to improve security in blockchain technologies.

helpnetsecurity.com

Step-DeepResearch Technical Report

Step-DeepResearch Technical Report

This content provides an in-depth overview of the Step-DeepResearch Technical Report. It invites readers to explore and discuss key findings, methods, and insights presented in the report, encouraging active engagement and collaboration. The page serves as a platform for exchanging ideas and fostering a deeper understanding of the research.

huggingface.co

Video

Silicon Minds, Human Hearts - Amit Bahree

Silicon Minds, Human Hearts - Amit Bahree

In this episode of "Silicon Minds, Human Hearts," Amit Bahree from Microsoft discusses the practical use and impact of generative AI in various domains. He shares how developers can quickly learn new programming languages, the importance of building trustworthy AI systems, and the balance between AI creativity and potential risks like "hallucinations." Amit also offers insights on responsible AI practices, favorite AI tools, and gives a glimpse into his book, *Generative AI in Action*. The conversation blends practical advice, honest reflections, and lighthearted moments to deepen understanding of AI’s role in shaping our world.

youtube.com

AI Show and Tell - Edge AI & Hyper Scale Data Labeling

AI Show and Tell - Edge AI & Hyper Scale Data Labeling

This video covers the topics of Edge AI and hyper-scale data labeling, presented during the AI Show and Tell event held in San Francisco on September 2, 2025. It offers insights into advancements in AI technologies, focusing on how Edge AI is transforming real-time processing and decision-making, while hyper-scale data labeling is revolutionizing the way data is prepared for AI models. Watch the full replay to explore these cutting-edge developments in AI.

youtube.com

Articles

Running Evals on a Bloated RAG Pipeline

Running Evals on a Bloated RAG Pipeline

This article focuses on the importance of running evaluations on a bloated Retrieval-Augmented Generation (RAG) pipeline to compare metrics across datasets and models effectively. It highlights the challenges of maintaining performance while balancing growing data demands and explores practical steps to streamline evaluations. The piece provides insights into refining workflows to ensure accuracy and efficiency in complex machine learning systems.

towardsdatascience.com

Prompt Engineering for Data Quality and Validation Checks

Prompt Engineering for Data Quality and Validation Checks

Prompt engineering can transform data quality and validation processes by framing questions in a way that mimics a data auditor's mindset. This method allows for faster and smarter quality assurance compared to traditional scripts, making it more efficient and adaptable. It highlights the potential of structuring prompts to optimize data verification tasks effectively.

kdnuggets.com

Upcoming Events

AgentCon - The AI Agents World Tour Continues in 2026

AgentCon - The AI Agents World Tour Continues in 2026

AgentCon continues into 2026 with the AI Agents World Tour—one-day, developer-focused conferences dedicated to autonomous AI agents. Building on a successful run of events, the tour expands to even more cities worldwide, from San Francisco to Singapore and beyond. Join leading engineers, researchers, and builders to explore cutting-edge agent architectures, real-world use cases, and emerging best practices. Connect with the global AI community and help shape the future of autonomous AI.

globalai.community

Code

Dangerous WebRAT malware now being spread by GitHub repositories

Dangerous WebRAT malware now being spread by GitHub repositories

Security researchers have identified 15 malicious repositories on GitHub being used to distribute the WebRAT malware. This malware is designed to compromise devices and steal sensitive information, putting users at significant risk. The discovery highlights the need for caution when downloading repositories from online platforms, as attackers are leveraging trusted sources to spread their malicious tools.

techradar.com

Shubhamsaboo/awesome-llm-apps: Collection of awesome LLM apps with AI Agents and RAG using OpenAI, Anthropic, Gemini and opensource models.

Shubhamsaboo/awesome-llm-apps: Collection of awesome LLM apps with AI Agents and RAG using OpenAI, Anthropic, Gemini and opensource models.

This GitHub repository features a curated collection of impressive LLM (Large Language Model) applications that utilize AI Agents and Retrieval-Augmented Generation (RAG). It highlights tools and resources powered by OpenAI, Anthropic, Gemini, and various open-source models. A valuable resource for exploring and building AI-driven applications.

github.com

Podcast

Machine Ethics Podcast episodes

Machine Ethics Podcast episodes

The Machine Ethics Podcast explores a wide range of topics surrounding AI and technology ethics. Featuring news, explanations, and interviews with experts like academics, authors, business leaders, and engineers, it covers areas such as artificial intelligence, machine learning, AGI, responsible AI, consciousness, and philosophy. It's an insightful resource for understanding the ethical implications of autonomous algorithms and modern technology.

open.spotify.com

>