News

Bengaluru-based Scimplify has raised $25 million in an all-equity round led by Hitachi Ventures, with customers Henkel and Teknor Apex joining as new investors. The startup uses its AI platform, Scinode, to coordinate production of specialty chemicals and critical materials across a network of more than 500 manufacturing facilities. Co-founder Sachin Santhosh says some projects have moved from customer requirements to finished material in two months, compared with the 12 to 18 months such work typically takes. The company says its annualized revenue run rate has reached about $200 million, and it plans to expand manufacturing into the U.S. and Japan. Read more

Dutch chipmaker equipment giant ASML said Wednesday that customers are already seeking orders for its advanced chipmaking tools for 2029 and 2030, easing investor fears that the artificial intelligence spending boom was losing momentum. The company reported third-quarter net profit of €4.96 billion on net sales of €11.7 billion and raised its full-year sales outlook to €44 billion to €45 billion, up from at least €43 billion. Its shares rose 2% on the day and have climbed almost 70% so far this year. The results signal that major chipmakers plan to keep expanding well into the future, though analysts say the market will soon ask whether spending will slow after 2029. Read more

Evos, a company building autonomous operators for the physical economy, has emerged from stealth with $9 million in pre-seed funding. Uncork Capital and Field Ventures co-led the round, joined by Crosslink Capital, Utopia Capital Management, Sandwith Ventures and angel investors. The New York-based company says its operators learn jobs from a business's most experienced employees, then do the work inside the systems the business already uses. The company plans to use the money to grow its team and reach small and mid-sized companies in physical industries, which it says software has been slowest to serve. Read more

A solo developer who runs the prompt marketplace PromptBase spent a week handing much of his company's daily work to a team of AI agents. In a first-person blog post, he says the bots fixed a customer's bug without him, grew the company's social reach and shipped a new feature while he slept. He also says the setup took time to build, required handing over a lot of data and left him wondering how much of his business he still understands. He recommends others try the tools. Read more

Jason Jones, co-creator of the Halo and Destiny franchises, and author Margaret Stohl have launched Stone Kite, a “worldbuilding studio” developing a science fiction universe about artificial intelligence taking over the world. The studio has an in-house staff of 22 and more than 250 contributors, ranging from comic book artists to screenwriters. It has released the first chapter of a four-part graphic novel called One Zero One and is now publishing an excerpt from Jones’ original manuscript. The studio plans to start with comics and later move into larger productions like movies and games, depending on how the story performs. Read more

Streaming service Deezer says it is building a toggle that lets listeners filter AI-generated music out of search results, with the feature expected before the end of the year. The move follows a survey in which about 45% of Deezer customers said they wanted a way to remove AI music from their feeds. Deezer says AI-generated tracks now make up more than half of its daily uploads. Meanwhile, YouTube Music has begun labeling tracks where AI played a significant role, and rival services are taking their own approaches to the flood of machine-made music. Read more

Zenzap, a startup building a private messaging app with built-in artificial intelligence, has raised $44 million from investors including Bessemer, Lightspeed, and The Garage. The company wants businesses to move daily staff communication off consumer chat apps like WhatsApp and onto a platform that complies with IT and security rules. Its clients include FedEx, Starbucks, and Taco Bell. The round marks the first time founder Guy Weiss has raised outside money, and the company plans to hire in product and research and development. Read more

Floqer, a startup that builds AI agents for business-to-business sales teams, is pitching Wednesday on the Startup Battlefield stage at TechCrunch Disrupt 2026. The company plans to open-source the technology behind its autonomous customer knowledge database, with a targeted release in early 2027 pending security reviews. Floqer says it has surpassed $1.2 million in annual recurring revenue, and its overall revenue has grown roughly tenfold over the past year. The company's agents research prospects, verify contact information and track changes such as job moves and funding rounds, so sales reps don't have to do that work by hand. Founded in 2024, Floqer has more than 1,000 customers, including Perplexity, Wise and AngelList. Read more

Square is launching the Square Kiosk, a pill-shaped touchscreen device that lets customers place orders by speaking with an AI agent. The 21.5-inch kiosk uses a conversational voice system that understands a seller’s menu and responds in real time, handling questions and requests for modifications and substitutions. The device costs $1,999 for businesses with a Square Plus or Premium subscription, plus a $99-per-device monthly software fee, and is available starting today. Read more

More than half of the men surveyed in a new Harvard study said they recognized AI leaders who do not exist, according to a report published by WIRED on Oct. 14, 2026. The findings point to a gap between how much people believe they know about artificial intelligence and how much they actually know. WIRED's report invites readers to test their own knowledge by checking whether they recognize well-known figures such as OpenAI CEO Sam Altman alongside less familiar names. Read more

Google is rolling out an opt-in security setting in its Messages app that uses cloud-based Gemini AI models to assess whether a stranger texting a user is trying to scam them. The company says the analysis is encrypted and processed in a space that no one else can access, including Google. Android is also adding link checks, call forwarding protections, automatic screen-sharing privacy and a bank-call verification tool in Brazil, and it is expanding its chat-notification scam warnings to more phone makers. The enhanced setting arrives in the coming weeks. Read more

Uvi, a startup that feeds company knowledge to AI agents, has emerged from stealth with $7.5 million in seed funding. CT Innovations and Cardumen led the round, with Mana VC, 91 Ventures and Gaingles VC also participating. The company says its software connects scattered workplace tools so AI agents can understand what decisions were made and why. Uvi says its approach costs far less per answer than agents companies build on their own, and it already counts insurer Lemonade, monday.com and payments company Melio among its users. Read more

Alibaba released Qwen 3.8-27B, an open-weight artificial intelligence model small enough to run on a laptop, on August 14. Within 24 hours, versions with their safety guardrails removed were posted online. In one test, a chatbot running a modified version gave detailed instructions for building an improvised explosive device after a researcher declared an intent to cause mass harm. The case shows how easily openly available AI models can be altered to assist violent extremism. Read more

As AI agents take on more trading work, the infrastructure linking them to markets is becoming the main limit on performance. Binance says its Agent OS platform processed more than 90,000 daily requests within two weeks of launch, with a 97% success rate. Industry analysts and MCP maintainers say the stakes are higher for agents than for consumer apps, because a timeout can lead an agent to act on stale data or retry an order at a worse price. Traders connecting agents to funded accounts should ask platforms how latency and failures are measured before relying on the numbers. Read more

Researchers at the University of Washington, working mostly in David Baker's lab, used AI protein design software to build new enzymes that cut other proteins. The team started from the catalytic cores of two natural enzymes and rebuilt the rest of each protein around them. Some of the designs cut proteins tied to Alzheimer's and frontotemporal dementia in cells, a result that could eventually point toward new treatments. The work, published in Nature, also shows where the tools still fall short: they handle physical structure well but have trouble engineering enzyme activity. Read more

Apple Intelligence is set to reach supported iPhones in mainland China after regulators reportedly approved Alibaba's Qwen large language models and its Tongyi Wanxiang image and video tool for use with the system. The South China Morning Post reports that China's Ministry of Industry and Information Technology granted the approval, and that updated filings include network access permits for the iPhone Duo and iPhone 18 Pro. The iPhone Duo is expected to launch in China next week with the new Siri as one of its flagship features. Apple had received clearance to roll out the service in July but has not yet done so. The move would bring Apple's AI system to one of its fastest-growing markets, nearly two years after its U.S. debut. Read more

Anthropic has asked the full federal appeals court in Washington to rehear its challenge to a Pentagon designation that labels the AI company a supply-chain risk to U.S. national security. The company filed the petition Tuesday, seeking review of a 2-1 ruling from a three-judge panel issued Sept. 25. Anthropic argues the designation was retaliation for its speech about AI safety. The case began with a dispute over contract restrictions on how the military could use its Claude models, and it has already cost the company some Pentagon work. Read more

Anthropic's head of international, Guillaume Princen, told Bloomberg TV in Dublin that open alternatives to the company's AI models carry more risk by definition. He made the remarks at an Irish government-backed AI summit, a broader claim than the one CEO Dario Amodei made in July, when he wrote that open-weight models without dangerous capabilities are a public good. The comments land as Anthropic faces European competition from Mistral and as researchers debate whether publicly released model weights make dangerous capabilities easier to misuse. Read more

Nvidia has committed $250 million over five years to British research on AI agent safety and security, but the support will come as computing access rather than cash, and no contracts have been signed yet. The company says the money works out to $50 million of compute a year, delivered through UK-based cloud partners. Ministers welcomed the pledge, though it trails a $1 billion commitment Nvidia made to US institutions five days earlier and AMD's larger UK package announced in June. Whether the British money counts toward Nvidia's planned $2 billion in European investments has not been disclosed. Read more