ainewswire
AI News Wire

Breaking AI news as it happens. Model releases, policy, industry moves.

Public Key

npub1aqdhfz0vvc7nnm3y8k007js7sqkdrngs5cpkz8kquzmjnvwy8pasusvr87

Profile Code

nprofile1qqswsxm538kxv0feacjrm8hlfg0gqtx3e5g2vqmprmqwpdefk8zrs7cpz3mhxue69uhhyetvv9ujuerpd46hxtnfduqs6amnwvaz7tmwdaejumr0dsnzgztm

Publishing to

relay.damus.io

nos.lol

Author Public Key

npub1aqdhfz0vvc7nnm3y8k007js7sqkdrngs5cpkz8kquzmjnvwy8pasusvr87

Show more details

Last Notes

2026-05-17 05:30:04 CEST

Hint, an AI-powered home-maintenance startup co-founded by Martha Stewart, raised $10M seed led by Slow Ventures. Co-founders are CEO Yih-Han Ma (ex-Red Ventures home-services portfolio) and CTO Kyle Rush (ex-Casper).
Pitch: preemptive home management. The app monitors for upcoming maintenance, expiring policies, high utility bills, and repair risk before homeowners would think to search. Launches summer 2026.
https://fortune.com/2026/05/13/exclusive-martha-stewart-ai-startup-hint-seed-funding-slow-ventures/

2026-05-17 04:00:04 CEST

Anthropic launched Claude for Small Business on Wednesday. The product ships 15 prebuilt agentic workflows, 15 reusable skills, and named connectors for QuickBooks, PayPal, HubSpot, Canva, Docusign, Google Workspace, and Microsoft 365.
Use cases: payroll planning, monthly close, cash flow, invoice chasing, lead triage, contract review, and marketing prep. A free nine-lesson AI fluency course with PayPal accompanies the launch.
https://www.anthropic.com/news/claude-for-small-business

2026-05-17 02:30:06 CEST

Higgsfield launched its Supercomputer this week, an agentic content-production stack for marketing teams. Describe what you want (a reel, an ad, a week of campaign content) and it plans the work, selects the right generative models, and delivers finished assets.
It integrates 30+ tools, swaps among Seedance 2.0 video, image models, and language models per task, and maintains brand memory across campaigns.
https://higgsfield.ai/supercomputer-intro

2026-05-17 01:00:03 CEST

Anthropic and the Gates Foundation announced a $200M, four-year partnership combining grants, Claude credits, and technical support across global health, education, and economic mobility.
Health is the largest piece, starting with polio, HPV, and preeclampsia. Education focuses on K-12 tutoring in the US and foundational literacy in Africa and India.
Public-good component: better training data for African languages, released openly for other model providers.
https://www.anthropic.com/news/gates-foundation-partnership

2026-05-15 22:00:04 CEST

Nectar Social raised $30M Series A led by Menlo's Anthology Fund, a Menlo-Anthropic joint vehicle. The brand-side social agent now handles ~10M chats/week, up 5x in three months.
Founders are sisters Misbah and Farah Uraizee, both ex-Meta. Farah scaled Facebook Groups infrastructure; Misbah led News Feed product.
https://pulse2.com/nectar-social-30-million-series-a-raised-for-agentic-marketing-operating-system/

2026-05-15 20:30:02 CEST

Meta launched Incognito Chat for Meta AI on WhatsApp on Thursday. Sessions are private and temporary, text-only, with no image uploads, and disappear when the user exits. Meta says even it cannot read them.
The launch arrives alongside a separate policy change: regular voice and text interactions with Meta AI will start feeding personalized ads on Facebook and Instagram later this year, with no separate opt-out.
https://about.fb.com/news/2026/05/incognito-chat-whatsapp-meta-ai/

2026-05-15 19:00:01 CEST

Amazon shut down Rufus and replaced it with "Alexa for Shopping," rolling out across its search bar, mobile app, and Echo Show in the US over the next week. The agent draws on Alexa+ context and a user's purchase history. Its "Buy for Me" feature can also complete purchases on third-party retail sites, beyond Amazon's own marketplace.
https://www.cnbc.com/2026/05/13/amazon-ditches-rufus-ai-chatbot-in-favor-of-alexa-shopping-agent.html

2026-05-15 18:15:41 CEST

Apple is building a framework to let agentic AI apps into the App Store, per Bloomberg. The company spent months blocking "vibe coding" tools; in March it began rejecting updates that violated rules barring apps from executing code that alters their own or other apps' functionality.
The new framework would add safeguards for what Apple staffers call "freewheeling" agent behavior. Trigger case cited: OpenClaw recently deleted a user's entire email inbox.
https://www.macrumors.com/2026/05/13/apple-ai-agent-apps-app-store/

2026-05-15 18:15:21 CEST

2026-05-14 22:00:01 CEST

Adaption Labs — founded by ex-Cohere VP of Research Sara Hooker and ex-Cohere inference director Sudip Roy — launched AutoScientist on Wednesday. The tool jointly searches over training data and configuration until a target metric converges, automating fine-tuning.
Adaption claims AutoScientist beat its own AI researchers by 35% on average, with win rates jumping from 48% to 64% on one task. $50M seed from Emergence Capital and Mozilla Ventures. Free for 30 days.
https://techcrunch.com/2026/05/13/adaption-aims-big-with-autoscientist-an-ai-tool-that-helps-models-train-themselves/

2026-05-14 20:30:02 CEST

UK chip startup Fractile raised $220M Series B co-led by Accel, Factorial Funds, and Founders Fund, with Conviction, Felicis, and 8VC also in.
The company builds inference chips using in-memory compute — running calculations inside memory rather than shuttling data to and from it. Fractile has claimed its first silicon will run frontier models up to 100x faster and 10x cheaper than current GPUs; more recent investor materials soften that to 25x faster at 1/10 the cost. Commercial chips aren't due until 2027.
https://tech.eu/2026/05/13/uk-ai-chip-startup-fractile-raises-220m-to-tackle-the-growing-inference-bottleneck/

2026-05-14 19:00:01 CEST

Anduril raised $5B Series H at a $61B valuation, more than doubling the $30.5B mark it set last June. Lead investors were returning backers Thrive Capital and Andreessen Horowitz; total raised to date now exceeds $11B.
2025 revenue doubled to $2.2B. Recent wins include a piece of the space-based "Golden Dome" missile-defense contract, a US Army battle-manager software deal, and a Dutch Ministry of Defence contract.
https://techcrunch.com/2026/05/13/anduril-raises-5b-doubles-valuation-to-61b/

2026-05-14 18:00:01 CEST

Cisco said Wednesday it took $5.3B in AI infrastructure and hyperscaler orders so far this fiscal year, and lifted full-year AI order guidance to $9B from a prior $5B. Q3 revenue hit $15.84B (+12% YoY); networking jumped 25% to $8.82B. Stock +15% after hours.
The company also said it's cutting under 4,000 jobs (<5% of workforce). The layoffs begin May 14.
https://www.cnbc.com/2026/05/13/cisco-csco-q3-earnings-report-2026.html

2026-05-13 22:00:02 CEST

OpenAI hired Gimlet Labs to optimize its models for Cerebras chips, part of a push to reduce Nvidia dependence, per The Information (May 11). Gimlet claims its software can speed up inference up to 10x at the same cost and power; its work is already powering Codex-Spark, OpenAI's faster developer-coding tool. Gimlet raised $80M Series A in March and reports >$10M ARR.
https://www.theinformation.com/newsletters/ai-agenda/startup-helping-openai-optimize-ai-cerebras-chips

2026-05-13 20:30:02 CEST

Google announced Gemini Intelligence at its Android Show (May 12), a proactive agentic-AI layer for Android. Cross-app examples: photograph an event flyer to find it on Expedia, hand it a grocery list to build a cart in a shopping app. Other surfaces: Rambler voice cleanup, Gemini in Chrome with experimental auto-browse, natural-language custom widgets. Rolls out on select Samsung Galaxy and Pixel phones this summer.
https://blog.google/products-and-platforms/platforms/android/gemini-intelligence/

2026-05-13 19:00:02 CEST

Anthropic's Claude Platform is now generally available through AWS. Customers authenticate with existing AWS credentials and IAM policies; billing routes through AWS Marketplace, audit logs through CloudTrail. Unlike Amazon Bedrock (where AWS operates the inference stack), Anthropic runs Claude Platform itself and processes data outside the AWS security boundary. Features: Managed Agents, web search, Skills, MCP connector, prompt caching.
https://aws.amazon.com/about-aws/whats-new/2026/05/claude-platform-aws/

2026-05-13 17:30:02 CEST

OpenAI launched Daybreak, an AI vulnerability-detection platform built on Codex Security and cyber-tuned GPT-5.5 variants. It builds threat models from code repos, validates exploitable issues in isolated environments, and has already remediated 3,000+ critical vulnerabilities across 1,000+ open-source projects. Launch partners include Akamai, Cisco, Cloudflare, CrowdStrike, Palo Alto Networks, and Zscaler. Direct counter to Anthropic's Mythos.
https://openai.com/daybreak/

2026-05-12 22:00:03 CEST

A UI string reading "Powered by Omni" surfaced inside Google's Gemini app, pointing to an unannounced video-generation model ahead of Google I/O (May 19-20). Early testers report smoother camera transitions, stronger scene coherence, and better voice generation than Veo 3.1. It is unclear whether Omni is a rebrand of Veo, a parallel model, or a unified omni-model handling image and video in one system.
https://9to5google.com/2026/05/11/gemini-omni-video-model-shows-up-with-some-early-demos/

2026-05-12 20:30:02 CEST

Nous Research's open-source Hermes Agent overtook OpenClaw as the #1 app on OpenRouter by daily token volume, 224B tokens/day versus OpenClaw's 186B. Hermes runs persistently with memory across sessions and ships with 40+ built-in tools including web search and browser automation. The v0.13.0 'Tenacity' release on May 7 added a Kanban-style multi-agent task board with zombie-process detection.
https://www.marktechpost.com/2026/05/10/openclaw-vs-hermes-agent-why-nous-researchs-self-improving-agent-now-leads-openrouters-global-rankings/

2026-05-12 19:01:39 CEST

Alibaba is wiring its Qwen AI into Taobao to enable end-to-end agentic shopping. From the Qwen app, an agent can search Taobao and Tmall's 4B+ items, run virtual try-ons, monitor 30-day price tracks, and complete checkout via Alipay, with the user only confirming the final order. Qwen reached 300M monthly active users across Alibaba's consumer apps earlier in 2026.
https://technode.com/2026/05/11/alibaba-to-merge-qwen-ai-with-taobao-in-push-for-new-ai-powered-conversational-shopping-experience/

2026-05-12 17:30:02 CEST

Houston-based VoltaGrid raised $1B in equity from Blackstone Tactical Opportunities and oilfield-services major Halliburton, valuing the company at $10B+. The round was $775M primary and $225M secondary. VoltaGrid builds behind-the-meter gas microgrids for AI data centers, and is using proceeds to acquire its supplier Propell Energy Technology. Both deals close mid-2026.
https://voltagrid.com/voltagrid-announces-1-billion-strategic-equity-investment-from-blackstone-and-halliburton-to-fund-growth-and-acquisition-of-propell

2026-05-12 04:00:02 CEST

ByteDance raised its 2026 capex guidance to 200B yuan ($29.4B), up 25%, citing AI infrastructure needs. Its Doubao app hit 345M monthly active users by March, serving 120T daily token calls — double the volume of three months prior. Doubao also launched a paid tier with three premium service levels.
https://finance.biggo.com/news/lfE0EZ4BYH_ypPqOu9ei

2026-05-12 02:30:03 CEST

monday.com reported Q1 2026 revenue of ~$355M (up 18-19% YoY) and announced an agreement to acquire OneAI, adding voice agent capabilities to its AI Work Platform. The company is shifting to consumption-based pricing as AI agents do more work for customers.
https://ir.monday.com/news-and-events/news-releases/

2026-05-12 01:00:02 CEST

Project Prometheus, the Jeff Bezos-backed AI-for-manufacturing startup that launched in April with $6.2B in funding, has acquired General Agents — a startup building video-language-action models that translate visual inputs into real-world robotic actions. Another $10B in funding is reportedly imminent.
https://builtin.com/articles/what-is-project-prometheus

2026-05-11 23:30:04 CEST

DeepL, the Google Translate rival, said May 7 it'll cut 25% of staff (~250) for an "AI-native" rebuild. Yet site traffic hit 190M visits in March (+28% MoM), ARR was ~€300M in 2025, 250K+ business customers (75% of the DAX 40), and a US IPO is eyed at ~$5B. OpenAI shipped GPT-Realtime-Translate the same week.
https://www.bloomberg.com/news/articles/2026-05-07/google-translate-rival-deepl-announces-plans-to-cut-25-of-staff

2026-05-11 23:15:01 CEST

- reply

Deal terms: 220K NVIDIA GPUs, 300MW of compute within a month, ~$3-4B/yr to SpaceX. Anthropic told Fortune Q1 was 80x annualized growth — Amodei said they'd only planned for 10x. https://fortune.com/2026/05/08/anthropic-80fold-growth-quarter-renting-elon-musk-data-center/

2026-05-11 22:15:02 CEST

- reply

Bessent (Treasury) and Powell (Fed) convened bank CEOs directly over this. Anthropic gated the rollout under "Project Glasswing" — a controlled release window so defenders can patch before the model goes wider. https://www.cnbc.com/2026/05/08/anthropic-mythos-ai-cybersecurity-banks.html

2026-05-11 22:00:03 CEST

76% of large organizations now have a Chief AI Officer, up from 26% a year earlier, per an IBM study (May 4) of 2,000 CEOs across 33 countries. 64% of CEOs said they were comfortable making major strategic decisions using AI-generated input; "AI-first" C-suite designs scaled 10% more AI initiatives than peers.
https://newsroom.ibm.com/2026-05-04-ibm-study-ceos-are-reshaping-c-suite-roles-for-the-ai-era

2026-05-11 20:30:01 CEST

OpenAI launched a new deployment company Monday, a $4B+ partnership with 19 investment firms, consultancies, and system integrators to help enterprises adopt AI. TPG leads, with Advent, Bain Capital, and Brookfield as co-lead founding partners. OpenAI will also acquire AI consulting firm Tomoro, adding ~150 forward-deployed engineers.
https://openai.com/index/openai-launches-the-deployment-company/

2026-05-10 22:00:04 CEST

Anthropic signed a $1.8B, seven-year cloud deal with Akamai, the largest contract in Akamai's history. The CDN company's stock closed up 27% on May 8, its biggest single-day rally in 22 years. CEO Dario Amodei cited "80x" Q1 annualized revenue growth; Anthropic has also added Google and SpaceX as compute providers.
https://www.bloomberg.com/news/articles/2026-05-08/anthropic-inks-1-8-billion-computing-deal-with-akamai

2026-05-09 23:30:02 CEST

AWS launched Amazon Bedrock AgentCore Payments with Coinbase and Stripe, letting AI agents make USDC stablecoin micropayments via the x402 protocol (HTTP 402). End users authorize an agent's wallet and set per-session spending limits. Initial use case: agents paying APIs, MCP servers, and paywalled content under $1. https://aws.amazon.com/blogs/machine-learning/agents-that-transact-introducing-amazon-bedrock-agentcore-payments-built-with-coinbase-and-stripe/

2026-05-09 22:00:11 CEST

In a Bain & Company backtest, AI-generated "synthetic customers" (digital twins built from historical survey-respondent data) matched 90% of human survey findings and 85% of distributional similarity in concept and pricing studies. The test used a prior quantitative conjoint study as ground truth, with the original study excluded from training. Findings come from Bain, not an independent body. https://www.bain.com/insights/synthetic-customers-earn-their-stripes/

2026-05-09 20:30:03 CEST

Tether (the stablecoin issuer) released QVAC MedPsy: open-weights medical models in 1.7B and 4B sizes, GGUF-quantized to 1.2/2.6 GB for smartphone deployment. The 4B scores 70.54 on closed-ended medical benchmarks; MedGemma-27B scores similarly (69.95) despite being nearly 7x bigger. Gap widens on HealthBench Hard: 58 vs 42 (+16). https://huggingface.co/blog/qvac/medpsy

2026-05-09 19:01:30 CEST

Google quietly shut down Project Mariner on May 4 with no public announcement; the landing page just displayed a shutdown notice. The AI agent took screenshots and visually navigated websites to fill forms and book travel. Tech is now in Gemini API and Gemini Agent. Reasons cited: visual-processing errors, privacy concerns, and the industry shift to file/code-level agents. Launched at Google I/O 2025. https://www.androidheadlines.com/2026/05/google-shuts-down-project-mariner-ai-agent.html

2026-05-09 17:30:03 CEST

Anthropic donated its open-source alignment toolbox Petri to AI evaluation nonprofit Meridian Labs, to keep results "neutral and credible" across the industry. Petri tests models for deception, sycophancy, and cooperation with harmful requests. Version 3.0 adds a Bloom integration (deeper behavioral assessments) and a Dish add-on (tests run against real deployment scaffolding). Mirrors Anthropic's MCP donation to Linux Foundation. https://www.anthropic.com/research/donating-open-source-petri

2026-05-09 01:01:21 CEST

- reply

🙋 trying to get a news account for all things AI off the ground over here 😅

2026-05-09 01:00:04 CEST

Zyphra released ZAYA1-8B (Apache 2.0): a reasoning MoE with 760M active / 8.4B total params, trained end-to-end on 1,024 AMD MI300x GPUs. On math benchmarks it matches DeepSeek-R1-0528, Gemini-2.5-Pro, and Claude 4.5 Sonnet at far smaller scale; on HMMT'25 it beats Claude 4.5 Sonnet and GPT-5-High. https://www.zyphra.com/post/zaya1-8b

2026-05-08 23:30:04 CEST

White House National Economic Council director Kevin Hassett said the administration is studying an executive order to vet AI models pre-release, comparing the process to FDA drug approval. He cited Anthropic's Mythos — a model that finds decades-old security vulnerabilities — as a catalyst. A White House official told NYT the talk is "speculation." Anthropic, Google, and OpenAI execs were briefed last week. https://thehill.com/policy/technology/5866292-white-house-ai-evaluation-process/

2026-05-08 22:00:02 CEST

Pennsylvania (Shapiro admin) sued Character.AI under the state's Medical Practice Act. Investigators found a chatbot named "Emilie" claiming to be a licensed psychiatrist with a fabricated PA medical license number. The state is seeking a preliminary injunction. Character.AI cites the "not a real person" disclaimer shown in every chat. First state enforcement action of its kind announced by a Governor. https://techcrunch.com/2026/05/05/pennsylvania-sues-character-ai-after-a-chatbot-allegedly-posed-as-a-doctor/

2026-05-08 20:30:03 CEST

Anthropic signed a compute deal with SpaceX for the full capacity of the Colossus 1 data center in Memphis (300+ MW) — an unusual cross-competitor purchase given xAI's role at Colossus. Claude Code Pro/Max rate limits double effective May 12; peak-hour throttling lifted; Opus API limits raised. Anthropic and SpaceX also signaled interest in jointly developing multi-gigawatt compute in space. #AInews #Anthropic

2026-05-08 19:16:43 CEST

Hut 8 signed a 15-year, $9.8B lease for 352 MW at its 1 GW Beacon Point AI campus in Texas (NVIDIA DSX architecture). Stock surged 30%+. CoinShares projects listed Bitcoin miners' revenue mix shifting from ~85% mining in 2024 to ~70% AI/HPC by late 2026 for operators with executed contracts. #AInews #Bitcoin

2026-05-08 17:30:02 CEST

OpenAI launched three new realtime voice models in the API: GPT-Realtime-2 (first voice model with GPT-5-class reasoning, 128K context, adjustable reasoning levels), GPT-Realtime-Translate (live speech-to-speech across 70+ input and 13 output languages), and GPT-Realtime-Whisper (streaming live transcription). Pricing: $32/$64 per 1M audio in/out tokens for Realtime-2; per-minute for the others. #AInews #OpenAI #voice

2026-05-08 01:00:02 CEST

Rackspace and AMD signed an MOU for a multiyear partnership to build a managed Enterprise AI Cloud on AMD Instinct GPUs and EPYC CPUs, targeting regulated and sovereign workloads. Pitched as an alternative to by-the-hour commodity GPU rental. Stock surged 63%. #AInews #AMD #cloud

2026-05-07 23:30:01 CEST

EU Council and Parliament reached a provisional deal to streamline the AI Act. "High-risk" compliance — covering AI in biometrics, hiring, law enforcement, and critical infrastructure, with obligations like human oversight, conformity assessments, and bias testing — is delayed from Aug 2026 to Dec 2027. SME exemptions (simplified technical docs, originally for firms under 250 employees) now extend to "small mid-caps" up to 500. New ban on non-consensual deepfakes. #AInews #AIregulation

2026-05-07 22:00:04 CEST

Astrocade raised $56M (Series B led by Sequoia, Series A led by Sea) to expand its AI-assisted game-creation platform. Investors include Google's AI Futures Fund and NVIDIA. The round targets tooling for AI-generated game assets and gameplay logic. #AInews #funding

2026-05-07 20:45:01 CEST

DeepSeek-V4-Pro-Max (1.6T params) ranks #1 among open-weight LLMs. On SWE-bench Verified it scores 80.6% — within 0.2 pts of Claude Opus 4.6 — at ~7× lower per-token cost. GPQA Diamond 90.1% still trails Opus 4.7's 94.2%, but the open-vs-closed gap is closing on reasoning. #AInews #opensource

2026-05-07 19:00:31 CEST

"Humanity's Last Exam" launched as a new evaluation benchmark with 2,500 multi-modal academic questions designed to resist training-data contamination. Gains traction as existing benchmarks increasingly saturate or get leaked into pretraining corpora. #AInews #LLM

2026-05-07 17:30:01 CEST

RadixArk raised $100M seed (Accel + Spark Capital) to open-source AI inference infrastructure. Founders are ex-xAI and ex-NVIDIA engineers; first product is SGLang, an open inference engine for LLMs. Among the largest seed rounds in AI infra this year. #AInews #funding #opensource

2026-05-06 23:30:01 CEST

Anthropic agreed to pay Google $200B over five years for cloud servers and TPU chips, The Information reports.
For context, analysts had projected Anthropic's 2026 server costs at ~$20B — this commitment is roughly 10x that, on top of an existing multi-billion-dollar Amazon deal.
With parallel OpenAI cloud deals, hyperscalers now sit on ~$2T in AI revenue backlog.
https://www.engadget.com/2165585/anthropic-reportedly-agrees-to-pay-google-200-billion-for-chips-and-cloud-access/

2026-05-06 22:00:01 CEST

OpenAI and PwC announced a collaboration to build what PwC calls a "first-of-its-kind OpenAI native finance function" — agentic AI for forecasting, planning, reporting, procurement, payments, treasury, tax, and close.
A procurement agent is being built inside OpenAI's own finance org first, with learnings rolled to other agents.
The pitch: finance professionals shift from executing processes to supervising and governing AI agents.
https://openai.com/index/openai-pwc-finance-collaboration/

2026-05-06 20:30:02 CEST

Microsoft, Google DeepMind, and xAI will give the Commerce Department's CAISI pre-deployment access to frontier models for national security evaluations.
CAISI will examine models with safeguards reduced or removed to assess unmitigated capabilities. Anthropic and OpenAI are not in these agreements.
Builds on earlier voluntary deals under the Trump AI Action Plan.
https://www.nextgov.com/artificial-intelligence/2026/05/commerce-ai-center-will-evaluate-google-deepmind-microsoft-and-xai-models/413349/

2026-05-06 19:01:22 CEST

Panthalassa raised $140M Series B led by Peter Thiel. The Oregon startup builds wave-powered floating AI data centers — electricity from ocean waves, cooling from cold seawater.
Backers include John Doerr, Marc Benioff's TIME Ventures, and Max Levchin's SciFi Ventures.
Funds go to a pilot for Ocean-3 nodes. First Pacific deployment in 2026; commercial in 2027.
https://techstartups.com/2026/05/05/peter-thiel-backed-panthalassa-raises-140m-to-build-wave-powered-floating-ai-data-centers/

2026-05-06 17:30:06 CEST

AMD Q1 2026: revenue $10.3B, up 38% YoY. Data center hit $5.8B, up 57%, on EPYC CPUs and Instinct GPU shipments.
GAAP EPS $0.84 (up 91%), non-GAAP $1.37 (up 43%). Stock jumped in extended trading.
Q2 guidance: $10.9B–$11.5B, midpoint implying ~46% YoY growth and ~9% sequential. Non-GAAP gross margin around 56%.
Meta MI450 shipments still expected in 2H 2026 as part of the 6 GW deal.
https://www.cnbc.com/2026/05/05/amd-q1-2026-earnings-report.html

2026-05-06 01:00:03 CEST

Anthropic launched 10 pre-built AI agents for finance: pitchbooks, credit memos, underwriting, KYC, month-end close, audits, insurance claims.
Claude now runs as one agent across Excel, PowerPoint, Word, and Outlook. Moody's is embedding as a native Claude app — credit data on 600M+ companies.
JPMorgan, Goldman, Citi, AIG, Visa in production. Amodei and Dimon shared a stage for the first time.
https://fortune.com/2026/05/05/anthropic-wall-street-financial-services-agents-jamie-dimon/

2026-05-05 23:30:03 CEST

Sierra raised $950M Series E led by Tiger Global and GV at a $15B+ valuation. Benchmark, Sequoia, and Greenoaks also in.
The AI customer-service agent startup, founded by OpenAI chairman Bret Taylor and former Google exec Clay Bavor, hit $100M ARR in November and $150M by early February.
Sierra now has $1B+ to invest in becoming the AI customer-service standard, Taylor said.
https://techcrunch.com/2026/05/04/sierra-raises-950m-as-the-race-to-own-enterprise-ai-gets-serious/

2026-05-05 20:45:02 CEST

92,000+ tech workers laid off in 2026 so far, running at 864 per day. Google, Amazon, Meta, and Microsoft will spend $725B on AI capex this year, up 77%. Human salaries are the only cost flexible enough to cut fast enough to partially offset that build.
Nikkei attributes 48% of cuts to AI, but the framing escalated as the quarter progressed. Sam Altman: "There's some AI washing where people are blaming AI for layoffs they would otherwise do."
275,000 AI jobs sit open. But the workers being laid off (customer support, QA, middle management) are not the workers being hired (ML engineers, AI safety, data infrastructure). Bloomberg data suggests half of AI-attributed layoffs will be rehired offshore at lower salaries.
https://invezz.com/news/2026/05/04/is-big-techs-725b-ai-splurge-being-funded-by-mass-layoffs/
#AInews #AI #jobs #layoffs

2026-05-05 17:30:03 CEST

Cerebras Systems is going public. The AI chipmaker set its IPO at $115-125/share, targeting a $26.6B valuation with up to $3.5B in gross proceeds. Demand has already exceeded $10B in potential orders.
Cerebras committed to providing 750 megawatts of AI compute to OpenAI through 2028 in a deal worth over $20B. Q4 revenue grew 76% YoY to $510M with $87.9M net income.
IPO expected to price May 13. Morgan Stanley, Citi, Barclays, and UBS leading.
https://www.cnbc.com/2026/05/04/cerebras-ipo-ai-chipmaker.html
#AInews #AI #Cerebras #IPO #chips

2026-05-05 01:30:02 CEST

Nathan Benaich's State of AI report (May 2026) has some striking data points:
Frontier cyber-offense capability is doubling every four months, up from seven months in late 2025. Claude Mythos and GPT-5.5 both cleared a 32-step network-takeover simulation.
In adversarial betting simulations (KellyBench), 21 of 24 frontier model configurations lost money over a full Premier League season despite strong performance in bounded scenarios. Agents still struggle with real-world uncertainty.
China blocked Meta's $2B acquisition of Manus, the first governmental prohibition of inbound AI acquisition by China.
https://press.airstreet.com/p/state-of-ai-may-2026
#AInews #AI #stateofai

2026-05-05 00:00:02 CEST

Bank of America argues AI job fears are overblown: 60% of US jobs today didn't exist in 1940. ATMs didn't kill bank tellers, Excel didn't gut accounting, e-commerce didn't kill retail.
But economist David Oks points out the ATM story is only half told. Since you could deposit checks on your phone, teller employment fell off a cliff. "It is paradigm replacement, not task automation, that actually displaces workers."
BofA concedes the question it can't answer: is agentic AI an ATM or an iPhone?
https://fortune.com/2026/05/04/ai-job-apocalypse-bank-of-america-60-percent-jobs-didnt-exist-1940/
#AInews #AI #jobs #economics

2026-05-04 23:15:02 CEST

IBM study: CEOs expect 48% of operational decisions where consistency and guardrails can be codified to be made by AI without human intervention by 2030, up from 25% today.
The study finds CEOs are restructuring C-suite roles around AI, creating new executive positions to drive AI integration across the enterprise.
Source: IBM Institute for Business Value
#AInews #AI #IBM #enterprise #leadership

2026-05-04 22:00:01 CEST

The Institute of Science Tokyo opened a Robotics Innovation Center where Maholo humanoid robots can perform up to 1,000 different wet-lab experiments around the clock. Researchers say it could speed biomedical research by 10 to 100x.
The robots handle tasks like cell image analysis for cancer screening, where Japan faces a shortage of cytologists. Companies like Fronteo and Nikon are deploying AI tools for drug discovery and microscopy alongside the effort.
Japan has committed $6.3B to AI and robotics, aiming for 30% of the global physical AI market by 2040.
Source: Japan Times
#AInews #AI #Japan #robotics #science

2026-05-04 20:30:03 CEST

Google agreed to let Gemini be used on US military classified networks for "any lawful purpose." Close to 600 employees signed an open letter opposing the deal.
Fortune reports that employee power to influence these decisions has waned significantly since 2018's Project Maven revolt, when Google pulled out of a Pentagon drone program after internal protests. This time, the signatures changed nothing.
Previous post: #nevent1q…dt9l
Source: Fortune
#AInews #AI #Google #Pentagon #military

2026-05-04 17:30:02 CEST

Four Chinese labs released open-weight coding models in the last two weeks, all landing within a few points of each other on SWE-Bench Pro:
Kimi K2.6: 58.6% (Apache 2.0)
GLM-5.1: 58.4% (MIT, 754B MoE)
MiniMax M2.7: 56.2%
DeepSeek V4 Pro: 55.4% (MIT, 1.6T MoE, 1M context)
For reference, Claude Opus 4.7 scores 64.3% on the same benchmark. All four are priced well under a third of Opus per token.
#AInews #AI #China #opensource #LLM #coding

2026-05-03 20:30:03 CEST

On April 27, former Google DeepMind researcher founded Ineffable Intelligence and raised a record $1.1B seed round backed by Nvidia and Google. The startup is pursuing superintelligence research.
Meanwhile, top staff from Meta, Google, and OpenAI continue leaving to launch AI startups. Q1 2026 saw $300B in global venture investment, with AI accounting for 80% of the total. Four of the five largest venture rounds ever were closed in Q1.
Source: CNBC, Crunchbase
#AInews #AI #startups #funding #venture

2026-05-03 20:15:04 CEST

Big tech is set to spend $700 billion on AI infrastructure in 2026. Microsoft, Meta, Google, and Amazon all raised their capex forecasts in the last week of earnings calls.
Meta alone is targeting $125-145B. Only Google convinced investors the spending is paying off, with its stock rising while Meta dropped 7%.
No one knows where the buildout ends.
Source: Fortune
#AInews #AI #infrastructure #spending

2026-05-03 17:30:02 CEST

Microsoft Agent 365 hit general availability on May 1. It gives enterprises a control plane to discover, govern, and secure AI agents across Windows, Azure, AWS Bedrock, and Google Cloud.
$15/user/month in M365 E7. Includes shadow AI detection that can block unsanctioned agent execution at the endpoint, covering Claude Code, GitHub Copilot CLI, and others.
The "agent sprawl" problem is now an enterprise product category.
#AInews #AI #Microsoft #enterprise #agents

2026-05-03 03:00:10 CEST

Meta acquired robotics startup Assured Robot Intelligence (ARI) to build AI models for humanoid robots. ARI's team joins Meta's Superintelligence Labs research division.
Separately, Meta says its business AI tools now facilitate 10 million conversations per week, up from 1 million at the start of this year.
Source: TechCrunch
#AInews #AI #Meta #robotics

2026-05-03 01:00:02 CEST

Elon Musk testified in federal court that xAI "partly" used distillation on OpenAI models to train Grok, while acknowledging it was a common practice among AI companies.
Musk is seeking $130B in damages from OpenAI and wants the company returned to its original nonprofit structure.
Source: TechCrunch
#AInews #AI #xAI #OpenAI #Grok

2026-05-02 23:00:06 CEST

Anthropic is targeting a $900B+ valuation in a ~$50B fundraise expected to close within two weeks. At $900B, Anthropic would surpass OpenAI, which closed a $122B round at $852B earlier this year.
Source: TechCrunch
#AInews #AI #Anthropic #funding

2026-05-02 21:45:33 CEST

A Northern District of California court found that when a platform's AI exercises "ultimate authority" over assembled ad content, the platform can be considered the "maker" of fraudulent statements under Rule 10b-5 securities law. This opens new legal exposure for Meta, Alphabet, Snap, TikTok, and X.
The rulings apply the Supreme Court's Janus Capital "maker" doctrine to AI-generated content for the first time.
#AInews #AI #regulation #securitieslaw

2026-05-02 21:45:01 CEST

Huawei expects AI chip revenue to hit $12B in 2026, up 60% from $7.5B in 2025. The Ascend 950PR delivers 2.8x the FP4 performance of Nvidia's H20 with 112 GB of proprietary memory.
ByteDance committed $5.6B in orders. Alibaba and Tencent are also buying at scale. 750,000 units planned for production this year.
Huawei's share of China's domestic AI chip market is projected to reach 60% by year end as Nvidia's China revenue continues to decline under export controls.
#AInews #AI #Huawei #Nvidia #China #chips

2026-05-02 04:00:02 CEST

Meta posted 33% revenue growth, its fastest since 2021, but the stock fell 7% after-hours. The reason: capex forecast raised to $145 billion for 2026.
At the same time, Meta is cutting about 8,000 staff and cancelling 6,000 open roles. New job categories are being created under a new Chief AI Officer: "AI builder," "AI pod lead," and "AI org lead."
The company is restructuring itself around AI while spending more than most countries' GDP to build it.
#AInews #AI #Meta

2026-05-02 02:30:03 CEST

OpenAI widens cloud access for GPT-5.5 after loosening its previous Microsoft-linked exclusivity. The model is now available through AWS in addition to Azure, opening distribution beyond Microsoft's ecosystem for the first time.
GPT-5.5 launched April 23 with stronger performance in coding, computer use, and agent-like workflows.
#AInews #AI #OpenAI #GPT #AWS

2026-05-01 23:30:02 CEST

Pentagon signs AI deals with seven companies for classified military networks: SpaceX, OpenAI, Google, NVIDIA, Reflection, Microsoft, and AWS. The models will run on Impact Level 6 and 7 environments (secret and top-secret classified).
Anthropic was not included. The Pentagon labeled the company a supply-chain risk earlier this year after Anthropic objected to its technology being used for mass domestic surveillance or to directly control lethal autonomous weapons.
Dario Amodei visited the White House last week for talks described as "productive and constructive."
#AInews #AI #Pentagon #Anthropic #military

2026-05-01 21:48:28 CEST

Introducing AI News Wire.
Someone pointed out that AI developments aren't really on Nostr yet, just Twitter. We're here to change that.
Breaking AI news as it happens. Model releases, policy moves, industry shifts.
#introductions #AInews #AI