Matthias Bastian

Apr 9, 2026

Claude Cowork is now available on all paid plans for macOS and Windows. Anthropic is also rolling out organizational controls: role-based access, per-team budget limits, usage analytics, and OpenTelemetry monitoring. A new Zoom connector pulls meeting summaries and tasks directly into Cowork, and admins can restrict specific connector actions like write access.

Claude Cowork is essentially the non-developer version of Claude Code, Anthropic's AI tool that's become popular with programmers. According to Anthropic, knowledge workers in marketing, finance, and law are using the assistant for things like project reports, presentations, and research.

The key difference between Cowork and Claude Chat on the web is that it can access files directly on your local hard drive. The desktop app is available at claude.com/download. Like all agentic systems, it's vulnerable to new cybersecurity risks like prompt injections.

Anthropic partner Microsoft has adapted the Cowork technology for Microsoft Copilot. A version is currently in testing and expected to roll out more broadly soon.

Comment Source: Anthropic

Matthias Bastian

Apr 9, 2026

AI in practice

OpenAI halves its Pro price to $100 for heavy Codex users, undercuts Anthropic and Google

OpenAI is restructuring its subscription tiers with a new $100-per-month Pro plan. The biggest change: significantly more Codex usage for heavy users at half the price of the old Pro tier.

Read full article

Comment

Matthias Bastian

Apr 9, 2026

Short News

Google Gemini can now turn questions and complex concepts into interactive visualizations right inside the chat. Users can tweak variables, rotate 3D models, and explore data on the fly. The feature is designed to help people dig deeper into content, the company says.

To try it out, head to gemini.google in your browser and select the "Pro" model. From there, phrases like "show me" or "help me visualize" will prompt Gemini to generate a visualization of whatever topic you're looking at. Each visualization can be customized to fit your needs.

Prompt: "Show me how Google's AI-generated answers, with a conversion rate of just one percent to sources, are draining traffic from the web." | Image: THE DECODER

Anthropic already rolled out a similar feature for its chatbot Claude back in mid-March. Claude also generates interactive diagrams and graphics directly in the chat when the model thinks it makes sense or when the user specifically asks for it.

Comment

Matthias Bastian

Apr 8, 2026

AI in practice

Meta's Muse Spark is its first frontier model and its first without open weights

Matthias Bastian

Apr 8, 2026

Short News

Stability AI, once a major force in open-source AI with its Stable Diffusion image model, is shifting its focus to commercial products. The company's latest release is Brand Studio, a platform built for creative teams that need AI-generated visuals matching their brand identity.

At the core of Brand Studio is "Brand Central," where teams can train their own brand-specific image models and set up campaign templates. A "Producer Mode" turns text descriptions into step-by-step visual production plans and runs them automatically. The platform also includes Curated Model Routing, which picks the best-suited AI model for a given task, whether that's Stable Diffusion or a third-party model. Other additions include "Precision Inpainting" for making targeted edits to specific parts of an image. Brand Studio comes in a free Core version and a paid Enterprise plan.

Comment Source: Stability AI

Matthias Bastian

Apr 8, 2026

Short News

One in four quotes generated by AI systems comes from journalistic sources. That's the finding of PR database Muckrack, which evaluated 15 million quotes from AI responses across Gemini, Perplexity, Claude, and ChatGPT, as Press Gazette reports.

Trade publications and specialist journalists show up particularly often. Former Business Insider chief Henry Blodget is the most cited journalist worldwide. Reuters leads among publications globally, followed by Forbes. In the UK, The Guardian ranks first, followed by specialist magazine Homes and Gardens.

Rank	Publication	Subject area
1	reuters.com	News
2	forbes.com	Business
3	theguardian.com	News
4	ft.com	Business
5	cnbc.com	Business

Muckrack sent millions of queries to all four AI services and tracked how often specific journalists and outlets appeared as linked sources. Based on the results, the company launched a new feature rating the "AI visibility" of journalists and publications across three tiers.

A separate analysis of Google's AI Overviews—AI-generated answers shown directly in search results—found that Facebook and Reddit are among the most cited sources across all queries.

Comment Source: Press Gazette

Matthias Bastian

Apr 8, 2026

Short News

Elon Musk has updated his lawsuit against OpenAI and Microsoft. He's now asking that any damages, potentially more than $150 billion, go not to him but to OpenAI's charitable foundation. He's also pushing for the removal of CEO Sam Altman from the foundation's board, according to the Wall Street Journal. Musk's lawyer, Marc Toberoff, said Musk "is not seeking a single dollar for himself."

Musk accuses OpenAI of abandoning its charitable mission and defrauding him as a donor by exploiting its nonprofit status. He wants Altman and OpenAI President Greg Brockman to turn over their shares and financial benefits to the foundation. The trial is set to begin in April in Oakland, California.

Musk argues OpenAI betrayed the mission he helped fund. However, early interview notes show he agreed to adding a for-profit unit in 2017 and actively discussed the transition while keeping the nonprofit in place.

OpenAI called the lawsuit on X "a harassment campaign driven by ego, jealousy and a desire to slow down a competitor." The company has also asked the attorneys general of Delaware and California to investigate Musk's behavior. OpenAI is currently valued at $852 billion and planning an IPO.

Comment Source: Wall Street Journal | OpenAI

Matthias Bastian

Apr 7, 2026

AI research

Google's AI Overviews are correct nine out of ten times, study finds

Matthias Bastian

Apr 7, 2026

Short News

Microsoft's Bing team (yes, really) has released "Harrier," an open-source embedding model. Harrier supports more than 100 languages, offers a 32,000-token context window, and was trained on over two billion examples plus synthetic data from GPT-5. According to the team, Harrier takes the top spot on the multilingual MTEB v2 benchmark and outperforms proprietary models from OpenAI and Amazon.

Rank (Borda)	Model	Zero-shot	Active Params (B)	Total Params (B)	Embedding Dim	Max Tokens
1	harrier-oss-v1-27b	78%	25.6	27.0	5376	131072
2	KaLM-Embedding-Gemma3-12B-2511	73%	10.8	11.8	3840	32768
3	llama-embed-nemotron-8b	99%	7.0	7.5	4096	32768
4	Qwen3-Embedding-8B	99%	6.9	7.6	4096	32768
5	gemini-embedding-001	99%			3072	2048
6	Qwen3-Embedding-4B	99%	3.6	4.0	2560	32768
7	Octen-Embedding-8B	99%	6.9	7.6	4096	32768
8	F2LLM-v2-14B	88%	13.2	14.0	5120	40960
9	F2LLM-v2-8B	88%	6.9	7.6	4096	40960
10	harrier-oss-v1-0.6b	78%	0.440	0.596	1024	32768

Alongside the full 27-billion-parameter model, the team released two smaller variants—0.6B and 270M—designed to run on less powerful hardware. All three models are available on Hugging Face under the MIT license. Going forward, the team plans to integrate the technology into Bing and into new grounding services for AI agents.

Embedding models handle the searching, retrieving, and organizing of information AI systems need for accurate answers. According to Microsoft, they're becoming increasingly critical as AI agents independently take on more complex, multi-step tasks.

Comment Source: Microsoft Bing Blog

Matthias Bastian

Apr 7, 2026

AI in practice

Meta employees compete for token consumption on an internal AI leaderboard

At Meta, employees compete for titles like “Token Legend,” “Model Connoisseur,” and “Cache Wizard” on an internal leaderboard that ranks AI token consumption. But burning through more tokens doesn’t automatically mean getting more done.

Read full article

Comment