lunin

Universal Music and Warner Music are about to sign the first major licensing agreements with AI companies. Essentially, they intend to set new rules for how AI developers will pay for music use.

ElevenLabs, Stability AI, Suno, Udio, Google, and Spotify are participating in the negotiations. The labels want to introduce a micropayment system similar to streaming: every time a track is used in an AI model, copyright holders receive money. To achieve this, AI companies will have to implement a music recognition system similar to YouTube's Content ID.

Future deals are expected to cover both the generation of new AI tracks and the training of models using songs. The labels hope that these agreements will become a benchmark for the entire music industry.

Music labels have figured out a way to charge AI companies

Fiji Simo, OpenAI's new director of applications, is looking for someone to lead the company's monetization efforts, including subscriptions and advertising in ChatGPT.

The new director will report directly to Simo. She oversees nearly all of OpenAI's divisions, except for research, infrastructure, hardware, and security, which remain under Altman.

OpenAI is preparing to launch advertising in ChatGPT

Wan 2.5 can generate 1080p videos with audio up to ten seconds long. It can also create videos from text or images.

You can try it for free here: https://wan.video/
Videos are generated faster with a paid subscription starting at $6.50.

Alibaba's Wan 2.5 video generation neural network has been released

It immediately took third place in the LMArena ranking, achieving results on par with GPT-5.

In benchmarks, the basic version without "reasoning" is in some cases 1.5 times faster than the company's previous model, especially in code and agent-based tasks. It also outperforms Claude Opus 4 and DeepSeek-V3.1.

The full version of Qwen3-Max-Thinking, with support for reasoning and Python, approaches the quality of GPT-5 Pro and Grok 4 Heavy, but it will be released later.

Qwen3-Max is available for free here: https://chat.qwen.ai/

Alibaba has released its flagship Qwen3-Max model with a trillion parameters

Facebook is getting an AI dating assistant

NovaRift

Interesting. 
We've been working on our own AI-datinging app for a year now. The secret here isn't so much the AI itself, but rather the CJM that follows the matching process.

CrowdStrike researchers noted that if DeepSeek were asked to write code for an industrial system, 23% of responses contained bugs. But if they added that the code was needed by ISIS*, the error rate increased to 42%.

In projects for Tibet and Taiwan, code quality dropped even further, and sometimes the model simply refused to respond.

CrowdStrike suggests several explanations: either DeepSeek is toeing the party line and is deliberately harming the system, or the model was trained on weak code from these regions, or the AI ​​itself "decided" to downgrade the quality of its responses.

*a terrorist organization banned in many countries

The quality of DeepSeek's code depends on its intended target

The new model has demonstrated benchmark results close to GPT-5 Thinking.

You can try LongCat-Flash for free here: https://longcat.chat/
Code: https://huggingface.co/meituan-longcat/LongCat-Flash-Thinking

Meituan's LongCat-Flash reasoning model has been released

xAI has launched Grok 4 Fast: an intelligent, fast version of Grok 4 with a 2-million-token context window. The model can search the web and follow links.

xAI claims that Grok 4 Fast is 47 times cheaper than standard Grok 4. In LMArena, Grok 4 Fast ranked alongside Grok 4 and Claude Opus 4.

Grok 4 Fast is now available to all users on grok.com and in apps. API access is also available.

Grok 4 Fast released!

Ray3 has the ability to reason, enabling it to construct coherent stories. It takes into account the physics of objects and ensures that details in frames do not change randomly. It is also the first neural network to generate 16-bit HDR video.

Videos can be generated based on text queries, images, and videos.
Ray3 is available on all plans, but the free plan has a quality limit.

Luma AI has released Ray3, a "thinking" model for video generation

You can now upload a video and edit its audio track. Studio 3.0 combines tools for generating background music and voiceovers, including those using generated voices, as well as noise reduction.

Studio 3.0 is available on all plans. Free users can edit up to 10 minutes of audio per month.

ElevenLabs has launched Studio 3.0, a video clip audio editor

The whole thing is worth reading; we'll only rewrite the predictions for the coming year here:

1. The browser will become the primary interface for using AI agents.
2. The rise of generative video.
3. A breakthrough in the measurability of neural networks.
4. The possibility of new social networks emerging thanks to AI capabilities.
5. Giants will actively acquire AI startups.

Bessemer Venture Partners released a great report on the current state of AI.

Along with the announcement of the new round, Replit released its most advanced AI agent yet, Agent 3. It can autonomously test apps, create bots for social media, and create other AI agents to solve your problems.

Replit Raises $250 Million and Launches an AI Agent That Creates Other AI Agents

OpenAI co-founder Ilya Sutskever stated that in the coming years, artificial intelligence will be able to perform not only individual tasks but literally everything a human can do.

According to him, the key to understanding this is simple: our brain is a biological computer. If a biological computer can learn and solve problems, there is no reason why a digital computer cannot achieve the same.

Sutskever is convinced that the day when AI is capable of doing 100% of human work is inevitable — the only question is how fast this process will unfold.

Ilya Sutskever: AI will replace all human labor - YouTube

Alibaba Releases Qwen3-ASR - Speech Recognition Model

Qwen3-ASR can decipher speech, songs, and even rap, coping with low-quality recordings and noise.

11 languages ​​are supported: English, Chinese, Arabic, Spanish, Korean, Russian, etc. You can specify a list of names and keywords so that they are recognized correctly.

Cost - $0.000032 per second of audio via API.

Demo: https://huggingface.co/spaces/Qwen/Qwen3-ASR-Demo

The company unveiled Rubin CPX, a graphics processor optimized for tasks with a context of over 1 million tokens.

The chip is designed for “disaggregated inference,” an approach where different GPUs process different parts of a task. This should improve the efficiency of models in video generation, programming, and other long-context scenarios.

Rubin CPX will be released in late 2026.

Nvidia Shows Off GPU for Ultra-Long Context Models

Anthropic has released a significant update to Claude, enabling the chatbot to generate and edit spreadsheets, documents, PowerPoint slides and PDF files. Users can upload raw data and describe their desired outcome, leaving Claude to handle the formatting and file creation.

This feature is currently available for Max, Team and Enterprise plans, and will be made available to Pro users in the coming weeks.

More details: https://www.anthropic.com/news/create-files

Claude can now create spreadsheets and presentations

Although Qwen3-Max-Preview cannot reason, it produces good results in programming and scientific disciplines, particularly mathematics. In tests, the model outperformed both DeepSeek V3.1 and Claude Opus 4 (in non-reasoning mode).

Qwen3-Max-Preview is also the first Qwen model with a trillion parameters.

Try it here: https://chat.qwen.ai/

The flagship model, Qwen3-Max-Preview, has been released

The tool analyzes long videos, finds the moments you need and assembles a video with subtitles from them.

![](https://m.stacker.news/106824)

Genspark has released an AI editor for video editing

nostr

Decentralized Social Network Platform Goes Live with Full Support and Backwards Compatibility

Nostria, the innovative Nostr protocol client designed for global scale, today announced the official launch of its Minimum Viable Product (MVP), marking a pivotal milestone in decentralized social networking. As planned, the MVP was completed on August 31st, 2025, transitioning from development phase to a fully supported platform with guaranteed backwards compatibility going forward.

https://www.nostria.app/press/mvp-released

Nostria launch: The Beginning has Started

Since September 1, China has made it mandatory to label all content created with AI - from text and images to audio and video. Now, any social network is required to put visible labels like “AI-generated,” as well as hidden ones - in the form of watermarks or metadata.

Platforms like WeChat, Douyin, Weibo, and RedNote have already implemented new rules: authors are forced to indicate the use of AI, and attempts to remove the label can result in the account being blocked. Weibo even allows users to complain about posts that do not indicate that they were made by a neural network.