Spotify Enters a New Phase with a Triple AI Play — Audiobooks, a Podcast-Rival App, and AI Summaries
On May 21, Spotify announced three new AI-powered features all at once. The one drawing the most attention is an **audiobook generation tool** built on ElevenLabs' voice synthesis technology. Authors and publishers can simply input text and generate an audiobook with natural-sounding narration. The company says new plans will roll out progressively through the end of the year.
The second is the release of a **desktop app** that competes directly with Google's NotebookLM. It features a "Personal Podcast" function that lets users create their own podcast content, and it is available as a research preview in more than 20 markets.
The third is an **AI-powered Podcast Q&A feature and Briefing generation**. Listeners can ask questions about a podcast's content and generate daily and weekly summary briefings based on their interests. Spotify's strategy of promoting "active consumption" of audio content is coming into sharp focus.
Cohere Open-Sources Command A+, Its Most Powerful Model
Canadian AI company Cohere has released **Command A+**, its most powerful language model to date, under the Apache 2.0 license. For Cohere, a company that has specialized in the enterprise market, giving away its strongest model for free marks a major shift in direction. It is drawing attention from the community as a new option in the open-source camp.
US Cyber Command Brings AI to Its Most Secret Networks — OpenAI and Google Models Deployed
US Cyber Command has stood up a task force to run AI models from OpenAI, Google, and others on **top-secret-level Pentagon and NSA networks**. In the background lies the fact that AI systems like Anthropic's Claude Mythos can find security vulnerabilities faster than even the best human hackers. Anthropic states that "comparable tools will become widely available within 6 to 24 months," and the very foundations of cybersecurity are shifting.
Google Gemini's AI Avatar Generates "Uncannily Real" Clones
A Wired reporter used Google Gemini's **AI Avatar Tool** to generate a video clone of themselves and reported that it was "uncannily similar" to them. Google positions this as "the future of creativity," but it has reignited debate over the ethical and social implications of anyone being able to generate lifelike clones of real people.
NVIDIA GTC Taipei Kicks Off at COMPUTEX — Jensen Huang Keynote on June 1
NVIDIA GTC Taipei at COMPUTEX is about to get underway. CEO Jensen Huang is scheduled to deliver a live keynote at the Taipei Music Center on June 1, covering a broad range of topics including AI factories, scaling infrastructure, agentic AI, and physical AI. The event is expected to be a key one for reading the AI industry's next direction.
Anthropic–Blackstone Joint Venture Acquires Fractional AI
The enterprise AI joint venture established by Anthropic and Blackstone has acquired **Fractional AI**. The move strengthens its AI-native enterprise services and signals that Anthropic's business is expanding beyond simply offering APIs into consulting and implementation support.
Google Tightens Limits on Its AI Pro Plan
Google has reportedly "nerfed" the usage limits of its AI Pro plan. The specific restrictions will be communicated to each user in due course, but the move is drawing attention as a sign that the era of "all-you-can-eat" AI subscriptions is coming to an end.
---