AI News

Follow the latest AI and automation news

21 Jul 20267 Safar 1448 AH

This Former Intel CEO Wants to Jumpstart Moore’s Law With Light

Pat Gelsinger, CEO of Intel, is striving to enhance artificial intelligence capabilities using tiny beams of light. This innovation aims to improve data processing speed and efficiency, facilitating the development of more powerful AI models. Through this technology, Gelsinger hopes to achieve significant performance leaps, thereby strengthening Intel's position in the AI market.

Models

21 Jul 20267 Safar 1448 AH

Google releases three new Gemini models — but no 3.5 Pro

Google has announced the launch of the Gemini 3.6 Flash and 3.5 Flash-Lite models, showcasing its ongoing efforts in AI development. These releases come at a time when there is increasing focus on AI strategies, while the absence of the Gemini 3.5 Pro model raises questions about the company's direction. This development could impact Google's position in the AI market, as experts anticipate that the lack of an advanced model may pose challenges in competing with other firms.

Agents

21 Jul 20267 Safar 1448 AH

Claude Cowork learns new skills through screen recordings and voice-over explanations

Anthropic's Claude Cowork app has launched a new feature that allows users to record their screens while completing tasks. Users can add voice commentary, enabling Claude to turn these recordings into reusable skills. This feature represents a significant step in enhancing user experience, as it allows for effective documentation of processes. By integrating voice comments with screen recordings, Claude can better understand context and develop skills based on the provided information.

💡

Introducing the ChatGPT for small business program

How might the ChatGPT program impact small business growth?

Governance

Anthropic’s $1.5 billion book piracy settlement approved by judge

21 Jul 20267 Safar 1448 AH

Governance

21 Jul 20267 Safar 1448 AH

Anthropic’s $1.5 billion book piracy settlement approved by judge

A federal judge has approved a $1.5 billion settlement with authors who accused Anthropic of training its AI models on copyrighted books. The settlement will provide authors with approximately $3,000 for each book allegedly pirated, marking a historic precedent in copyright cases. This ruling represents the largest known copyright recovery in history, highlighting the increasing tensions between AI innovation and copyright protection.

Models

Google ships three new Gemini Flash models but its frontier 3.5 Pro remains lost in training

21 Jul 20267 Safar 1448 AH

Models

21 Jul 20267 Safar 1448 AH

Google ships three new Gemini Flash models but its frontier 3.5 Pro remains lost in training

Google has announced the launch of three new models in the Gemini series, including the more efficient 3.6 Flash, which uses 65% fewer tokens. The 3.6 Flash model features advanced technologies, making it an attractive option for governments and select partners, alongside a cybersecurity model available exclusively to them. However, the flagship model, Gemini 3.5 Pro, remains missing, placing Google in a challenging competitive position against OpenAI, Anthropic, and other Chinese labs advancing in this field.

Governance

A Sneaky Hacking Tool Targeting AI Infrastructure Is Lurking in Victims’ Blind Spots

21 Jul 20267 Safar 1448 AH

Governance

21 Jul 20267 Safar 1448 AH

A Sneaky Hacking Tool Targeting AI Infrastructure Is Lurking in Victims’ Blind Spots

A new type of malware has been discovered that can infiltrate AI coding systems. This malware is capable of stealing login data and sensitive information, posing a significant cybersecurity threat. The malware operates by burrowing into systems, where it can activate a so-called “death switch” that destroys files and locks out legitimate users. This technique reflects an evolution in cyberattack methods, necessitating immediate responses from companies.

Agents

Google’s Gemini 3.6 Flash targets enterprise agent token costs

21 Jul 20267 Safar 1448 AH

Agents

21 Jul 20267 Safar 1448 AH

Google’s Gemini 3.6 Flash targets enterprise agent token costs

Google has announced the launch of Gemini 3.6 Flash and 3.5 Flash-Lite, designed to reduce latency and token costs for AI agents. The new 3.6 Flash model shows significant improvements, cutting output tokens by 17% compared to its predecessor, enhancing efficiency in production environments. Through these models, Google aims to meet the needs of teams building background agents rather than chat interfaces, focusing on productivity and cost reduction. The company reports reductions of up to 65% in token usage in specific tests, showcasing the model's effectiveness in handling complex tasks.

Models

Alibaba's Qwen-Image-3.0 renders full infographic grids and readable ten-pixel text in a single pass

21 Jul 20267 Safar 1448 AH

Models

21 Jul 20267 Safar 1448 AH

Alibaba's Qwen-Image-3.0 renders full infographic grids and readable ten-pixel text in a single pass

Alibaba's Qwen team has launched Qwen-Image-3.0, an image generator that accepts prompts of up to 4,500 tokens. The new tool can render legible text as small as ten pixels and supports twelve languages natively, enhancing its usability across various applications. However, while it can create complex layouts like infographics and articles, its practical value remains unclear when the output is a pixel image rather than an editable format.

Infrastructure

NVIDIA Vera Rubin Driving Performance Per Watt, Lowest Token Cost for Partners Worldwide

21 Jul 20267 Safar 1448 AH

Infrastructure

21 Jul 20267 Safar 1448 AH

NVIDIA Vera Rubin Driving Performance Per Watt, Lowest Token Cost for Partners Worldwide

NVIDIA has announced the launch of Vera Rubin NVL72, ramping up production in data centers such as CoreWeave, Google Cloud, Microsoft Azure, and Oracle Cloud. This system spans over 350 factory sites across 30 countries, making it the largest integrated supply chain to meet compute demand. The Vera Rubin platform features an integrated design combining seven chips and five rack trays, enhancing performance while reducing operational costs. At the core of this system is the NVIDIA Vera CPU, which delivers superior performance compared to competing designs, making it ideal for agentic workloads.