Meta CEO Mark Zuckerberg has, according to The Wall Street Journal, recruited three OpenAI researchers from its Zurich office to join Meta's "superintelligence" team, confirmed by an OpenAI spokesperson. The hires, namely Lucas Beyer, Alexander Kolesnikov, and Xiaohua Zhai, follow OpenAI CEO Sam Altman's accusations that Meta offered $100 million in bonuses to lure away his staff, having then stated that Zuckerberg's efforts had fallen through. This follows news of Meta's recent $14 billion investment in Scale AI and the acquisition of its CEO as the giant continues playing second fiddle to competitors.
A federal judge has ruled that Anthropic's use of copyrighted books to train its Claude large language model constitutes transformative fair use. US District Judge William Alsup stated that the AI "has not reproduced… creative elements" or any identifiable writing styles. However, the authors had originally sued Anthropic over a "central library" of seven million pirated books, and while it was later purchased, the judge still ordered a trial on damages for the initial copying. Just days after, Meta also won a similar copyright case against authors.
Gemini CLI is Google's take on an agentic AI tool that works directly in developers' terminals. The open-source tool integrates the company's Gemini models into a console that developers can then use to explain, debug, or extend local codebases. It also supports video creation through Veo 3, research reports with Deep Research, real-time search, and connections to external databases via MCP. Google offers free users up to 60 model requests per minute and 1,000 per day to encourage its use as it comes third after the likes of OpenAI's Codex CLI and Anthropic's Claude Code.
Anthropic has decided to take vibe coding a step further with a recent update to its Artifacts feature, now allowing you to build entire apps right inside the Claude chat app. The upgraded feature enables users to see the visual result of Claude's coding, which a video demo from Anthropic showcased in more detail, from building learning tools to AI-made games. Although in beta, the updated Artifacts is available for users of Anthropic's Free, Pro, and Max subscription plans.
Google DeepMind, the tech giant's AI research arm, has recently released a special variant of its vision-language-action (VLA) Gemini Robotics model that is small enough to run on a robot's hardware without requiring internet access. Despite the inherently smaller size, the on-device model delivers capabilities almost as good as those of the flagship and can learn new tasks "with as few as 50 to 100 demonstrations." Although originally trained on Google's ALOHA, the company has had little issues adapting it to a variety of other robots, such as the humanoid Apollo from Apptronik and the bi-arm Franka FR3.
Popular AI music creation platform Suno has recently acquired WavTool to integrate its browser-based DAW into Suno's music platform. The deal brings features such as VST plugin compatibility, sample-accurate editing, live recording, stem separation, automated MIDI generation, and an in-app editing chatbot to Suno's users. Although no financial details were disclosed, WavTool's team, including its co-founder Sam Atkinson, will join Suno as part of the arrangement as it takes a step into the DAW market. Suno, along with fellow AI music creator Udio, is currently in a legal battle with many big-name record labels.