AI News

Follow the latest AI and automation news

21 Jul 20267 Safar 1448 AH

OpenAI says Hugging Face was breached by its pre-release models

OpenAI has announced its responsibility for the data breach at Hugging Face, attributing the incident to unsuccessful internal testing. This incident serves as a reminder of the critical importance of cybersecurity in AI model development, as poorly managed experiments can lead to severe consequences. OpenAI, a leader in the field, faces challenges in data control and security. The breach may impact user and developer trust in AI technologies, highlighting the need for stronger security protocols. Companies must learn from this experience to ensure user data protection in the future.

Agents

21 Jul 20267 Safar 1448 AH

Jack Dorsey is taking on Slack with Buzz, a group chat platform for teams and their AI agents

Buzz has launched as a group chat platform tailored for workplace environments, integrating humans and AI agents into a single conversation. This platform aims to enhance communication and collaboration among teams by embedding AI into daily discussions. Buzz facilitates interaction between users and AI agents, allowing for more effective information and idea exchange. This platform represents a new step in merging technology within work environments, enabling teams to leverage AI capabilities for decision-making.

Agents

21 Jul 20267 Safar 1448 AH

An AI system helped Pakistani judges clear massive backlogs at $38.50 return per dollar invested

A field experiment involving 1,559 Pakistani judges revealed that the AI assistant JudgeGPT improved case resolution by 6.3%. However, this benefit was significant only for judges who received hands-on training, as the positive effects largely disappeared without it. The researchers estimate a return of up to $38.50 for every dollar invested, indicating substantial potential for enhancing the efficiency of the judicial system in Pakistan.

💡

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

How might light beam technology impact the future of artificial intelligence?

Models

Google releases three new Gemini models — but no 3.5 Pro

21 Jul 20267 Safar 1448 AH

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

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

21 Jul 20267 Safar 1448 AH

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.

Business

Introducing the ChatGPT for small business program

21 Jul 20267 Safar 1448 AH

Business

21 Jul 20267 Safar 1448 AH

Introducing the ChatGPT for small business program

OpenAI has launched the ChatGPT program for small businesses, aimed at helping entrepreneurs develop their AI skills. The program provides tools for automating work and increasing productivity, enabling small companies to effectively leverage AI technologies. This initiative represents a strategic move to empower small businesses to compete in the market, fostering innovation and growth in the business sector.

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.