AI Pulse: Daily Digest — March 26, 2026

The Verge

OpenAI has officially discontinued Sora, its video-generation tool launched in late 2024, ending a high-profile licensing deal with Disney that was reportedly worth $1 billion. Reports indicate Disney was blindsided by the decision and that no money had yet changed hands. The shutdown marks a significant strategic retreat for OpenAI in the consumer AI media space.

The Verge

Sen. Adam Schiff is drafting legislation to enshrine Anthropic's self-imposed limits on AI use in autonomous weapons and mass surveillance into federal law, ensuring humans retain final authority over life-and-death decisions. The move comes amid an ongoing dispute between Anthropic and the Pentagon over military applications of its Claude model. A separate bill from Sen. Elissa Slotkin would further restrict the Defense Department's AI deployment authority.

The Verge

Meta is cutting hundreds of jobs across recruiting, social media, sales, and its Reality Labs division as the company doubles down on AI investment. The layoffs signal a deliberate reallocation of headcount and capital toward generative AI initiatives rather than hardware and metaverse bets. This continues a broader industry pattern of large tech firms trimming non-AI workforces to fund AI buildouts.

TechCrunch

Senator Bernie Sanders and Representative Alexandria Ocasio-Cortez have introduced companion legislation that would freeze all new data center construction until Congress enacts comprehensive AI regulation. The bill reflects growing progressive concern about the environmental and social costs of unchecked AI infrastructure expansion. It is unlikely to pass the current Congress but signals a sharpening political fault line around AI's physical footprint.

The Verge

The White House has named Meta's Mark Zuckerberg, Nvidia's Jensen Huang, Oracle's Larry Ellison, and Google co-founder Sergey Brin as the first members of the reconstituted President's Council of Advisors on Science and Technology. The 13-member panel will advise on AI policy at a moment when the administration is shaping the regulatory and investment landscape for the technology. The appointments underscore how deeply Silicon Valley's biggest AI players are embedded in federal policymaking.

Ars Technica

Google has unveiled TurboQuant, a new AI memory compression algorithm that can shrink the working memory requirements of large language models by up to sixfold while preserving output quality — a key differentiator from prior compression methods. The technique is currently a research result and has not yet been deployed in production systems. If it scales, TurboQuant could meaningfully reduce the hardware costs and energy demands of running frontier AI models.

The Verge

Incoming Disney CEO Josh D'Amaro is facing twin crises in his first week: the collapse of the company's $1 billion OpenAI Sora partnership following that product's shutdown, and mounting criticism over AI-generated content on Disney+. The setbacks raise serious questions about Disney's strategy of leaning into generative AI and virtual-world experiences as growth drivers. The situation illustrates the reputational and financial risks of high-profile AI partnerships built on products still in flux.

TechCrunch

AI legal technology startup Harvey has confirmed a new funding round that values the company at $11 billion, with Sequoia Capital leading and Andreessen Horowitz, Kleiner Perkins, and Elad Gil also participating. The valuation reflects surging investor appetite for vertical AI applications in high-value professional services. Harvey's rapid ascent makes it one of the most richly valued enterprise AI startups outside of the foundational model tier.

The Verge

Anthropic has introduced an auto mode for Claude Code that allows the AI to make permissions-level decisions independently, reducing the need for constant user approval while avoiding the risks of unconstrained autonomy. The feature is designed as a middle ground for developers who want faster AI-assisted coding without surrendering meaningful oversight. It reflects a broader industry push to make agentic AI tools more practical without abandoning safety guardrails.

TechCrunch

AI meeting assistant Granola has closed a $125 million funding round, vaulting its valuation from $250 million to $1.5 billion as it expands beyond transcription into a broader enterprise AI application with agent support. The six-fold valuation jump in a single round underscores how quickly the market is rewarding AI productivity tools that demonstrate enterprise traction. The company has also addressed earlier user complaints about limited AI agent integration.

Google AI Blog

Google has released Lyria 3 Pro, an upgraded music generation model capable of producing tracks up to three minutes long — six times the previous 30-second limit — with greater customization and integration across Google's suite of creative and enterprise products. The model is also available to developers in paid preview through the Gemini API and Google AI Studio. The launch positions Google as a serious competitor in the rapidly growing AI music generation market.

TechCrunch

Reddit is rolling out a new system that will flag accounts exhibiting automated or suspicious behavior and require them to verify they are operated by a human. CEO Steve Huffman announced the initiative alongside a new labeling system for accounts that are legitimately registered as bots. The move comes as AI-generated content and bot-driven manipulation continue to threaten the authenticity of user-generated platforms.

TechCrunch

Kleiner Perkins has raised $3.5 billion across two new funds — $1 billion targeting early-stage AI startups and $2.5 billion for late-stage growth investments — signaling a full-throated commitment to AI as the firm's defining thesis. The raise is one of the largest in Kleiner's recent history and reflects the continued flood of institutional capital into the AI sector. The firm is already an investor in Harvey, among other AI portfolio companies.

The Verge

Arm has unveiled its first self-designed CPU — the Arm AGI CPU — marking a historic departure from its decades-long model of licensing chip designs to third parties. Meta will be the launch customer, deploying the chip in its AI data centers later this year for inference workloads including AI agents. The move signals Arm's ambition to capture more value in the AI infrastructure stack beyond IP licensing.

TechCrunch

New research from Anthropic finds that while AI has not yet triggered widespread job displacement, a measurable skills gap is emerging between workers who use AI tools effectively and those who do not. Early data suggests experienced AI users are gaining a compounding productivity advantage, raising concerns about future workforce inequality. The findings add empirical weight to debates about whether AI will democratize opportunity or concentrate it.

TechCrunch

Spotify is piloting a new feature that gives artists greater control over which tracks appear under their name on the platform, targeting the growing problem of AI-generated music being falsely associated with real musicians. The tool is a direct response to the proliferation of so-called "AI slop" — low-quality, algorithmically produced content that can dilute an artist's catalog and mislead listeners. It represents one of the music industry's most concrete platform-level responses to AI content fraud.