Researchers at the MIT Media Lab looked at the cognitive costs of using LLMs like ChatGPT in everyday writing tasks. Their 200-page report, "Your Brain on ChatGPT," focused on how AI tools affect ...

Bloomberg reports that Elon Musk's AI startup xAI is currently burning through about $1 billion every month. The main drivers are steep infrastructure and chip costs for building its Grok chatbot.

The new "Image-to-Video" feature lets users turn any Midjourney image into a short animated clip. Animation is handled through a new "Animate" button in the Midjourney web interface. Users can choose ...

OpenAI has launched "ChatGPT Record", a feature in the macOS desktop app that lets Pro, Team, Enterprise, and Edu users record, transcribe, and summarize audio. Each session allows up to 120 minutes, ...

OpenAI has significantly updated ChatGPT's search feature: it now handles longer contexts, better follows instructions, answers complex questions with several parallel searches, and allows users to ...

TikTok has introduced three new AI tools—"Image to Video," "Text to Video," and "Showcase Products"—that enable advertisers to create video content in a more cost-effective way. These tools ...

A new website called "The OpenAI Files" launched on June 18, 2025, compiles internal documents and criticism of OpenAI's leadership, strategy, and corporate culture. It raises concerns about whether ...

Earlier plans for a dedicated OpenAI chip factory with TSMC are on hold. TSMC CEO C.C. Wei called OpenAI CEO Sam Altman's proposals "too aggressive," citing concerns about facility utilization rates.

OpenAI has released GPT-4.5 as a "research preview," describing it as their largest and best model for chat. The new model is initially available to ChatGPT Pro users and developers, with Plus and ...

Claude 3.7 Sonnet builds on an upgraded version of Claude 3.5 Sonnet but adds a significant new feature: an "Extended Thinking" mode that allows the model to reason through its responses. According to ...

ChatGPT appears to be inconsistent in how it handles requests for Studio Ghibli-style images. While testing for this article, both outcomes occurred - some attempts triggered copyright-related warning ...

Researchers at EPFL have discovered a significant security flaw in leading AI language models: Rephrasing malicious queries in the past tense often allows users to bypass protective measures and ...