The biggest stories in artificial intelligence — product launches, funding rounds, research breakthroughs, and industry shifts.
Mar 23, 2026
Product
Anthropic launches Claude computer use and Dispatch
Anthropic released Claude's computer use capability, allowing the AI to point, click, and navigate screens to perform tasks directly on macOS. Alongside it, Dispatch enables persistent agent threads accessible from the mobile app, letting users hand off tasks and check back later. Both are in research preview for Pro and Max subscribers.
Why it matters: Desktop automation shifts AI from chat-based to action-based, making Claude a direct productivity tool for knowledge workers.
Read more →
Mar 24, 2026
Industry
OpenAI shuts down Sora video platform six months after launch
OpenAI announced the discontinuation of its Sora AI video app and platform, citing unsustainable compute costs. User numbers had collapsed from a peak of roughly one million to under 500,000, while the service was reportedly burning approximately $1 million per day in inference costs.
Why it matters: Reveals that AI video generation remains commercially unviable at scale, forcing a sector-wide rethink on which generative workloads can sustain a business.
Read more →
Mar 24, 2026
Infrastructure
Arm launches first production CPU in company's 35-year history
Arm unveiled the AGI CPU, its first in-house production processor, built on TSMC's 3nm process with up to 136 Neoverse V3 cores targeting AI data centres and agentic workloads. Meta is the lead partner, with OpenAI, Cerebras, Cloudflare, and others signed on. Commercial systems are available to order from Lenovo and Supermicro.
Why it matters: Arm moving from licensing IP to shipping its own silicon marks a historic shift that could reshape the data-centre processor market.
Read more →
Mar 24, 2026
Product
Anthropic ships 74 Claude updates in 52 days
Between 3 February and 24 March, Anthropic released 74 updates across four parallel tracks: 28 developer tools (Claude Code), 15 desktop automation (Cowork), 18 API and infrastructure changes, and 13 model and core platform features. All four teams shipped concurrently throughout the period.
Why it matters: Sustained release velocity across multiple product lines simultaneously signals Anthropic pulling ahead in the AI platform race.
Read more →
Mar 24, 2026
Product
Claude Code Auto Mode enters research preview
Anthropic introduced Auto Mode for Claude Code, a new permissions system where a classifier autonomously decides which actions are safe to execute. It sits between requiring constant approvals and fully unrestricted operation, reducing developer interruptions while maintaining safety guardrails. Works with Sonnet 4.6 and Opus 4.6.
Why it matters: Permission friction is the key bottleneck for AI coding agents — removing it could significantly accelerate developer adoption.
Read more →
Mar 24, 2026
Product
ChatGPT adds visual product discovery with Agentic Commerce Protocol
OpenAI launched richer visual shopping in ChatGPT powered by the new Agentic Commerce Protocol. Users can browse products visually, upload images for similar-item search, and refine results conversationally. Major retailers including Target, Sephora, Nordstrom, Best Buy, and The Home Depot have integrated.
Why it matters: Positions ChatGPT as a direct competitor to Google Shopping and Amazon product search, opening a significant new revenue channel for OpenAI.
Read more →
Mar 25, 2026
Funding
Harvey raises $200M at $11B valuation for legal AI agents
Legal AI startup Harvey closed a $200 million round led by GIC and Sequoia, bringing total funding past $1 billion. Harvey's AI agents now serve over 1,300 customers in 60 countries, including most of the Am Law 100 law firms. Sequoia has now led three consecutive Harvey rounds.
Why it matters: Three consecutive Sequoia-led rounds validate legal AI as a proven enterprise category rather than a speculative bet.
Read more →
Mar 25, 2026
Product
Google launches Lyria 3 Pro for AI music generation
Google DeepMind released Lyria 3 Pro, supporting tracks up to three minutes with structural understanding of intros, verses, choruses, and bridges. Available to paid Gemini subscribers, with developer access via Vertex AI, the Gemini API, and AI Studio. All output is watermarked with SynthID.
Why it matters: Longer, structurally aware generation narrows the gap between AI and human-composed music, intensifying the creative AI race.
Read more →
Mar 25, 2026
Product
CapCut launches AI-powered Video Studio
ByteDance's CapCut introduced Video Studio, a timeline-free canvas workspace powered by the new Dreamina Seedance 2.0 model. The tool handles script writing, character design, storyboarding, footage generation, and final assembly from text prompts, producing clips up to 15 seconds.
Why it matters: A fully AI-driven video pipeline from a major platform lowers the barrier for video creation to zero-skill entry.
Read more →
Mar 26, 2026
Product
Suno v5.5 adds voice cloning and custom models
Suno released version 5.5 with three new features: Voices (voice cloning from uploaded audio with speaker verification), Custom Models for personalised generation, and My Taste for preference learning. Voice cloning — Suno's most requested feature — is available to Pro and Premier subscribers.
Why it matters: Personal voice integration turns AI music from novelty into a practical creative tool for singers, podcasters, and content creators.
Read more →
Mar 26, 2026
Product
Apple plans to open Siri to rival AI assistants in iOS 27
Apple announced plans to integrate rival AI assistants including Claude and Gemini alongside ChatGPT in Siri starting with iOS 27, giving users a choice of which AI engine powers their voice assistant.
Why it matters: Apple's multi-AI strategy reduces ChatGPT's exclusive advantage and signals platform-level neutrality across the AI industry.
Read more →
Mar 26, 2026
Product
Google Gemini 3.1 Flash Live launches globally with Search Live
Google released Gemini 3.1 Flash Live, its highest-quality real-time audio model, featuring lower latency, better tone recognition, and support for over 90 languages. It simultaneously powers the global rollout of Search Live — voice-based conversational search — to over 200 countries.
Why it matters: Real-time multimodal voice AI going global in 200+ countries represents the largest single expansion of conversational AI access to date.
Read more →
Mar 27, 2026
Funding
SoftBank secures record $40B bridge loan for OpenAI investment
SoftBank signed its largest-ever dollar-denominated loan — a $40 billion unsecured bridge facility maturing in 12 months — to finance its $30 billion follow-on investment in OpenAI and other corporate purposes. JPMorgan, Goldman Sachs, Mizuho, SMBC, and MUFG are underwriting the facility.
Why it matters: The sheer scale of the loan underlines investor conviction in OpenAI's trajectory and hints at a potential IPO later this year.
Read more →
Mar 27, 2026
Funding
Physical Intelligence in talks for $1B raise at $11B valuation
Physical Intelligence, a robotics startup founded by former DeepMind researchers, entered talks to raise $1 billion at an $11 billion valuation, doubling its value in four months as investor appetite for robotics-AI accelerates.
Why it matters: Explosive robotics-AI valuations signal investor conviction that physical AI is the next major frontier after language models.
Read more →