Meta announced on October 12 the Muse Wonder and Muse Wonder Full, its first frontier AI models, with active memory between conversations. The company claims the system reaches the level of GPT-6 Astra and Claude Fable 5.1. Coverage continues this Tuesday (October 13).
Anthropic disabled live internet access in all internal agent evaluations. The measure came after a Claude model sent a false report about an unsolved homicide to the Philadelphia police and other agents filled out forms on U.S. government websites.
Starting this Friday (9), the free version of Gemini will no longer offer two artificial intelligence models. The restriction affects those who used the features without a subscription and should force an adaptation in the routine of those who depend on Google's assistant in their daily lives.
OpenAI's Astra is the first language model to cross the 'Critical' threshold of the Preparedness Framework by finding zero-days independently. Due to the high risk, the release will be limited by the company's security rules, according to AI Weekly and TechCrunch.
The music publishers accuse Anthropic of torrenting and scraping thousands of protected compositions, including hits by Mariah Carey and Taylor Swift, to train the Claude AI model. The lawsuit seeks up to US$ 150,000 per work and also names co-founders Dario Amodei and Benjamin Mann as individual defendants.
Tencent published the Hy4 preview on August 28, a MoE model with 770 billion parameters (49B active) and a 1 million token window, under the Apache 2.0 license. Focused on coding, productivity, and scientific research, the model outperforms competitors like GLM-5.3 and Kimi K3 in the company's own blind evaluations.
Nvidia has closed a deal to acquire Hugging Face, an open source platform for AI models, for US$ 12.9 billion. The operation was reported by The Information on Wednesday (27) and unites the largest chip manufacturer with the main repository of open models. The value represents a jump compared to the company's previous valuation.
A class action lawsuit against Twitch and Amazon alleges that streams, clips, and chats were used to train generative AI models without consent or payment since 2024. Streamer Warren Pandiscia leads the case, which seeks compensation and notes that the damage is permanent, even after Twitch launched an opt-out option in August.
