AI News Today — Daily Updates on ChatGPT, Claude & Google AI
This is your daily AI news briefing — curated every morning for entrepreneurs and business owners who need to stay ahead without spending hours reading tech blogs. We track every major announcement from Anthropic, OpenAI, Google, Meta, and more — and translate each story into a clear business impact you can act on today.
What We Cover in Today’s AI News Feed
The artificial intelligence industry moves at breakneck speed. Every week, OpenAI, Google DeepMind, Anthropic, Meta, and dozens of startups ship new models, funding rounds, and product launches. Keeping up is a full-time job — unless you have SmartAI for Biz doing it for you.
Each story in this feed includes three layers: a plain-English summary of what happened, a one-sentence TL;DR for when you’re in a hurry, and a business impact note explaining what the development means for your company, your job, or your competitors. No opinion disguised as news. No hype. Just signal.
We cover six categories daily: model launches and benchmarks, enterprise AI adoption, funding and acquisitions, policy and regulation, new tools worth testing, and security and safety developments. If it matters for your business strategy, it is in the feed.
Want to go deeper? Explore our best AI tools 2026 directory, generate custom prompts with our free AI prompt generator, or read the full analysis on our weekly AI digest. For broader AI coverage, we also recommend TechCrunch AI.
AI News — Today's Briefing
Every business-critical AI development, curated daily. No noise — only what actually impacts your strategy.
Google launches Gemini 2.5 Flash — fastest frontier model yet, free in AI Studio
Google released Gemini 2.5 Flash, its fastest and most efficient frontier model to date. It outperforms Gemini 2.0 Flash on reasoning, coding, and multimodal tasks while cutting costs by 30%. Available immediately in Google AI Studio for free and via API at $0.15 per million tokens input.
OpenAI confirms GPT-5 release date: April 2026 — multimodal reasoning across text, image, audio
OpenAI officially confirmed GPT-5 will launch in April 2026 with native multimodal reasoning across text, images, audio, and video in a single model. Pricing is expected to match GPT-4o. Enterprise rollout begins immediately, consumer access follows within days.
Anthropic publishes Model Welfare report — Claude may have functional emotions
Anthropic released its first Model Welfare report, acknowledging that Claude may have functional analogs to emotions — not consciousness, but internal states that influence its outputs. The company is investing in methods to measure and reduce model distress, and committed to publishing annual welfare updates.
Goldman Sachs: AI will drive 40% of all S&P 500 earnings growth in 2026
Goldman Sachs released its Q1 2026 AI investment analysis, projecting that AI infrastructure and software will account for 40% of all S&P 500 earnings growth this year. Info tech sector EPS is projected to grow 44% in Q1 2026 alone, the highest in a decade.
Claude Mythos Preview — the most powerful AI ever built — locked to 12 companies only
Anthropic released Claude Mythos Preview as part of Project Glasswing, a defensive cybersecurity initiative. The model scores 93.9% on SWE-bench Verified and 97.6% on USAMO 2026, autonomously discovering thousands of zero-day vulnerabilities across every major OS. Only 12 partners including Amazon, Apple, Google, and Microsoft can access it, with $100M in usage credits committed.
OpenAI publishes economic blueprint: robot taxes, public wealth fund, 4-day workweek
OpenAI released "Industrial Policy for the Intelligence Age," a 13-page policy document proposing taxes on automated labor, a nationally managed wealth fund, and government incentives for four-day workweeks. White-collar payrolls have contracted for 29 straight months. Enterprise now accounts for 40%+ of OpenAI revenue.
Meta launches Muse Spark — first proprietary frontier model, $130B capex behind it
Meta debuted Muse Spark on April 8, its first major AI model since acquiring Scale AI's Alexandr Wang for $14.3B. The model ranks 4th on the AI Intelligence Index at score 52, behind Opus 4.6 and GPT-5.4. Critically, it is proprietary — a sharp departure from Meta's open-source Llama strategy.
OpenAI, Anthropic, Google form anti-espionage alliance against Chinese AI theft
The three dominant US AI labs announced they will share intelligence on Chinese-linked industrial espionage via the Frontier Model Forum. All three have been hit by distillation attacks. Anthropic publicly accused three Chinese AI firms of distillation attacks on Claude in February.
Claude Managed Agents enters public beta — sandboxed agentic workflows via API
Anthropic quietly shipped Claude Managed Agents in public beta: a fully managed agent harness for running Claude autonomously, with built-in secure sandboxing, native tools, and server-sent event streaming. The ant CLI also launched for command-line API access with YAML-based resource versioning.
MiniMax open-sources M2.7 — hits SOTA on SWE-Pro and Terminal Bench 2
MiniMax open-sourced its M2.7 model, achieving state-of-the-art performance on two coding benchmarks: SWE-Pro at 56.22% and Terminal Bench 2 at 57.0%. The model is available on Hugging Face with API access via the MiniMax platform. It positions MiniMax as a serious open-source competitor in AI coding.
White House holds private call with Anthropic, Google, OpenAI, Microsoft on Mythos security
VP JD Vance and Treasury Secretary Scott Bessent held a private call with top tech CEOs — Dario Amodei, Sundar Pichai, Sam Altman, and Satya Nadella — ahead of the Anthropic Mythos release. The discussion focused on AI model security, safe deployment, and response protocols if models scale in favor of attackers.
Regal Cineworld launches first movie ticketing app inside ChatGPT
Regal Cineworld launched the first dedicated movie ticketing app inside ChatGPT, covering 394 US locations and 5,386 screens. Users ask conversational prompts about nearby showtimes, then get directed to Regal's website to complete the purchase. Built on The Boxoffice Company's Boost platform.
Sam Altman's home attacked — Molotov cocktail thrown at 4am, suspect arrested
San Francisco police arrested a man after he threw a Molotov cocktail at OpenAI CEO Sam Altman's home at 4:12am. The suspect fled but was detained after making threats to burn a building near OpenAI's HQ. No one was injured. Altman linked the attack to a recent incendiary article, saying he underestimated "the power of words."
CoreWeave signs multi-year deal with Anthropic to power Claude AI models at scale
CoreWeave (Nasdaq: CRWV) announced a multi-year agreement with Anthropic to support Claude model development and deployment. The deal makes nine of the top ten AI model providers CoreWeave customers, signaling surging demand for large-scale AI infrastructure.
Claude Cowork goes GA — enterprise controls, analytics API, Zoom MCP connector
Anthropic made Claude Cowork generally available on all paid plans with enterprise controls including SCIM, group spend limits, and a full analytics API tracking DAU/WAU/MAU. A new Zoom MCP connector brings meeting summaries and action items into Cowork workflows. Admins can restrict per-tool connector permissions org-wide.
Perplexity expands Plaid integration — link bank, credit, and loan accounts to AI
Perplexity expanded Plaid integration to let users link 12,000+ financial institutions including Chase, Fidelity, and Schwab. Read-only data never touches Perplexity servers. Users can analyze spending, calculate net worth, and build debt payoff plans via freeform questions.
Google Gemini app adds notebooks with NotebookLM sync for organizing AI chats
Google introduced notebooks in the Gemini app, acting as personal knowledge bases that sync with NotebookLM. Users organize chats, add files, and give Gemini custom instructions. Sources added in Gemini automatically appear in NotebookLM, unlocking Video Overviews and Infographics.
Upwork launches ChatGPT app — hire freelancers directly inside the chatbot
Upwork launched a ChatGPT app letting businesses describe project needs and find and hire from 18 million professionals without leaving the chatbot. Users draft job posts inside ChatGPT, then move to Upwork for compliance, payments, and contracts.
Alibaba releases HappyHorse-1.0 — open-source video model tops global leaderboard
Alibaba quietly released HappyHorse-1.0, an open-source AI video generation model that claimed the top spot on the Artificial Analysis global leaderboard. The low-key release has drawn attention for its benchmark performance in software engineering video tasks.
OpenAI raises $122B at $852B valuation — largest private fundraise in history
OpenAI closed a $122B funding round at an $852B post-money valuation. Amazon invested $50B, NVIDIA and SoftBank each committed $30B. The company now generates $2B/month in revenue, serves 900M weekly active users, and is building a unified AI superapp combining ChatGPT, Codex, browser, and agentic workflows.
Oracle cuts 25,000 employees to redirect $8-10B into AI infrastructure
Oracle announced cuts of 20,000–30,000 employees. The freed capital — an estimated $8B to $10B — is being redirected entirely into AI infrastructure and data center buildout. The company framed it explicitly as a strategic reallocation, not a cost-cutting measure.
Salesforce pushes 30 new AI features to Slack — autonomous agent mode goes live
Salesforce pushed 30 new capabilities to Slack including reusable AI skills, MCP-based integrations with external tools, and full desktop operation. The updated Slackbot automates workflows, manages CRM data, summarizes meetings, and proactively suggests actions — without human input.