AI news, 13 August: CoreWeave results supercharge AI infrastructure stocks

CoreWeave and peers drive AI stock rally. Strong second-quarter results from AI cloud provider CoreWeave lifted markets on 12 August, according to Reuters. The company reported revenue of about $2.58 billion, more than double the year-earlier figure and slightly above expectations, while raising its full-year outlook and capital-spending plans. Its shares jumped roughly 19 percent. Fellow AI infrastructure names also surged: Nebius Group climbed about 34 percent after beating estimates, Super Micro Computer rose around 19 percent on an upbeat forecast, and chip-related stocks including Nvidia gained. The moves underscored continued heavy demand for the computing power behind generative AI, even as questions linger about long-term returns on the massive buildout.

Cognition AI in talks at $40 billion valuation. Coding-focused startup Cognition is in early discussions with investors for a new funding round that could value it at $40 billion or more, Bloomberg reported, citing people familiar with the matter. That would mark a sharp rise from its roughly $26 billion valuation in a $1 billion round less than three months earlier. The company’s annualized revenue run-rate is approaching $1 billion. The talks highlight intense investor appetite for AI tools that help write and improve software, though no deal is finalized and terms could change.

Spotify to label AI-generated artists. Spotify announced it will add “AI Persona” badges to profiles of fully AI-generated artists starting in mid-September and will exclude their music from editorial playlists and algorithmic recommendations by default. Artists can self-disclose, and the platform will also review popular accounts; incorrect labels can be appealed. The move aims to give listeners clearer transparency amid a flood of synthetic music while still allowing AI as a creative tool for human artists. Labels appear on profiles, search, and track listings.

German group files criminal complaint over Meta AI glasses. Digital-rights organization HateAid lodged a criminal complaint in Germany against Meta, Ray-Ban owner EssilorLuxottica units, and several retailers over the sale of Ray-Ban Meta smart glasses, Reuters reported. The group argues the devices enable inconspicuous filming that violates German privacy rules barring covert recording tools. Prosecutors confirmed receipt and will make a preliminary assessment. German regulators have previously flagged connected glasses but note that clearly signaled recording functions are generally permitted; the case adds to European scrutiny of wearable AI cameras.

Anthropic adds watermarks to Claude output. Anthropic confirmed it is embedding invisible, machine-readable watermarks in text (and provenance metadata in supported files) generated by Claude models launched on or after 2 August 2026, aligning with EU AI Act transparency requirements. The marks are designed to travel with copied or lightly edited text. Older models are being updated. The step helps platforms, educators, and others detect AI-generated content and reflects growing industry and regulatory focus on provenance as synthetic media becomes widespread.