AI News — Friday, July 17, 2026

9
Google Vids Now Lets Users Star in Their Own AI-Generated Videos

Google Vids introduces new features allowing users to create and star in personalized AI-generated videos, enhancing its creative capabilities.

TechCrunchproduct
9
The Agent Security Gap: Over Half of Enterprises Report AI Agent Incidents

A new report reveals that 54% of enterprises have experienced AI agent-related security incidents, with many still allowing agents to share credentials, highlighting critical security vulnerabilities.

VentureBeatindustry
8
Roblox Launches AI-Powered Game Creation Feature on Mobile App

Roblox integrates an AI-powered game creation tool directly into its mobile application, democratizing game development for a wider audience.

TechCrunchproduct
8
Google's AI Mode Expands to Link and Interact with Select Apps

Google's AI Mode now allows users to connect and interact with various third-party applications, significantly expanding its utility and integration within the Google ecosystem.

TechCrunchproduct
8
Ring-Zero: Scaling Zero RL to a Trillion Parameters for Emergent Reasoning

Researchers propose Ring-Zero, a method to scale Zero Reinforcement Learning to a trillion parameters, aiming to achieve emergent reasoning capabilities in AI models.

Hugging Faceresearch
7
Enterprises Face AI Compute Gap, Buying Infrastructure Faster Than Measuring Costs

Enterprises are rapidly acquiring AI infrastructure without adequate cost measurement, leading to a significant 'AI compute gap' and potential financial inefficiencies.

VentureBeatindustry
7
Harness Handbook: Making Evolving Agent Harnesses Readable, Navigable, and Editable

This paper introduces a 'Harness Handbook' to improve the readability, navigability, and editability of evolving agent harnesses, crucial for advanced AI agent development.

Hugging Faceresearch
7
Boogu-Image-0.1: Boosting Open-Source Unified Multimodal Understanding and Generation

Boogu-Image-0.1 is presented as a new open-source model aimed at significantly advancing unified multimodal understanding and generation capabilities.

Hugging Faceopen-source
7
AI Data Centers and the Concentration of Wealth

Bruce Schneier discusses how the proliferation of AI data centers contributes to the concentration of wealth, raising concerns about economic and power imbalances.

Lobste.rsindustry
6
Function-Aware Fill-in-the-Middle for Coding Agent Foundation Models

A new mid-training technique called 'Function-Aware Fill-in-the-Middle' is introduced to enhance coding agent foundation models, improving their ability to complete code based on functional context.

Hugging Faceresearch
6
Why Teens Deserve Access to Safe AI

OpenAI publishes an article advocating for safe and responsible AI access for teenagers, emphasizing the importance of tailored safeguards and educational resources.

OpenAI Blogindustry
6
Every AI-Generated Line of Code Is a Small Loan — And Eventually, You Have to Pay It Back

An opinion piece argues that AI-generated code, while convenient, incurs technical debt that developers will eventually need to address, highlighting the long-term implications of AI assistance in coding.

Dev.toindustry
6
KnowAct-GUIClaw: A Personal GUI Assistant with Self-Evolving Memory and Skill

KnowAct-GUIClaw is introduced as a personal GUI assistant designed to deeply understand user interactions and perfectly execute tasks, featuring self-evolving memory and skill acquisition.

Hugging Faceresearch
5
What is an "Agentic Harness," Actually?

This article provides a clear explanation of what an 'agentic harness' is, detailing its role and significance in the development and deployment of AI agents.

Dev.toindustry
5
OvisOCR2 Technical Report

The OvisOCR2 Technical Report details advancements in optical character recognition, showcasing improved performance and capabilities for document processing.

Hugging Faceresearch