最新 AI 资讯
Ex-Spotify employees raise $10M to bring the AI behind its recommendations to e-commerce
Sidd Motwani, Ian Anderson and Shivaditya Sinha spent years building the behavioral intelligence infrastructure behind Spotify’s recommendation engine. Called Vector AI, the system is designed to predict a person’s intent and next actions instead of relying only on their past behavior. It powers about 90% of Spotify’s recommendations to its 800 million users. Now, the three are bringing a similar system to e-commerce with their new startup,Malachyte. The company on Thursday said it had raised $10 million in seed funding to scale distribution and hire more product and commercial leaders. Malachyte was formed from the belief that most online stores treat shoppers the same way: Personalization is largely dictated by historical purchases, demographic segmentation, or logged-in customer profiles. That means first-time visitors often see the same generic storefront as everyone else, while existing shoppers receive recommendations based primarily on what they bought previously rather than what they need today. The startup wants to change that by building real-time, intent-aware shopping experiences. Its platform uses what it calls “two-headed Vector AI” to predict what product a shopper wants next, learn their general taste, and fine-tune continuously based on what they do in real time. “[Our] system starts forming before the first click, using the context available the moment the page loads. Within a single session, we build a real read on both preferences and what someone is trying to accomplish right now,” Motwani, Malachyte’s CEO, told TechCrunch. “A search for ‘heavy-duty boot’ followed by two clicks on steel-toed boots is enough to move work pants and gloves up the page and push dress shoes down, with no account or history required. Every additional action sharpens the profile, so the experience gets more relevant the longer someone stays, and again on their next visit.” Motwani argues that retailers already possess their most valuable source of customer intelligence, but rarely take advantage of it in real time. “Every hover, click, scroll, search refinement and add-to-cart is a signal, and most systems either never act on it in the moment or aggregate it into a segment overnight. We read it continuously, so each action makes the user’s vector more confident about both preference and current intent,” he added. He also believes contextual signals remain significantly underutilized. “A phone visitor at 11 p.m. from an email link is in a different state of mind than the same person on a laptop mid-morning, and most systems treat them identically,” Motwani said. The company has been developing and testing its technology since 2024, and worked with more than 20 enterprise customers across travel, grocery and retail before ultimately focusing on e-commerce. Its platform first went live in the fall of 2025 with Fun.com. Since June 2026, it has been generally available to Shopify merchants through a native integration, while larger retailers can integrate the technology through its API. Looking ahead, Motwani says the bigger opportunity is in bringing merchandising and marketing together around the same understanding of customer behavior. The funding round was co-led by Bessemer Venture Partners and Gradient, with participation from Harpoon Ventures.
View
Amid legal battles, Suno says it will start watermarking songs
Suno, the service that allows users to create AI-generated songs, today announced new tools to mark tracks made on its platform, limit downloads, and update community guidelines to prevent copycat songs. The changes come asSunofaces numerous lawsuits from labels andartist bodies. Ina blog post, co-founder and CEO Mikey Shulman shared core principles and said that the platform wants to promote original creation while enabling more people to make music with its AI tools. One of the key points of contention involved users uploading AI-generated songs on other streaming platforms and gaming the system to earn revenue. Suno said that now it will use audio watermarking and fingerprinting to prevent misuse on other streaming platforms. It’s not clear if Suno will use an existing system likeGoogle’s Synth IDor adopt a new one, and the company did not say when contacted by TechCrunch about this. The startup also signed an agreement with lyrics provider Musixmatch to use its Sentinel system for copyright detection, the blog post said. “These tools are designed to be durable and resistant to tampering, without affecting the listening experience. They are also not intended to pass judgment on whether a song is good, meaningful, or sufficiently human,” said Shulman in the blog. “Ultimately, we believe it should be up to artists and platforms to decide what they want to disclose. Our role is to build tools that give them transparency options and make it easier to collaborate across the industry.” The company added that it plans to add a new download policy to bar mass distribution on streaming platforms, but declined to provide details on the record. Suno has also changed itscommunity guidelinesto explicitly prohibit “deceptive audio presented as real” and “using a real person’s voice or likeness without permission” to prevent copycats. The company, which raised$400 million in a Series D funding round in June, is fighting legal battles on many fronts. The startup is in a lawsuit withthe Universal Music Group (UMG) and Sony Music Groupin a case coordinated by the Recording Industry Association of America (RIAA). Late last month, a German court ruled in favor of a government-mandated licensing agency, GEMA, and saidSuno was breaking copyright rules. Separately, 404 Media reported thatSuno experienced a data breachin November 2025, which revealed that the platform scraped YouTube, Deezer, and Genius to train its models. Later, data breach notification service Have I Been Pwned said thatthe data breach affected 55 million users. The startup now faces a class action lawsuit in Massachusetts, which alleges thatthe company overlooked security measures to focus on profit.
OpenAI says Apple’s own security practices undermine its trade secrets case
OpenAI’s motion to dismiss Apple’s trade secrets lawsuit, along with newly filed exhibits, reveals the company’s legal defense strategy. Rather than focusing on whether former Apple employees working at OpenAI had accessed certain information, the AI company argues that Apple’s own security practices and offboarding procedures weaken its claim that the information qualifies as legally protected “trade secrets.” Apple’s complaint,filed in July, accuses OpenAI of orchestrating a scheme to obtain confidential hardware information from former Apple engineers. This week, Appleasked the court to expedite discovery, saying its internal investigation indicates some additional former employees may have participated in, or witnessed, the alleged theft of trade secrets. In its motion, OpenAI argues that Apple allowed employees to use personal iCloud accounts for work and failed to properly revoke access after they left the company. It also submitted text message records showing that an Apple manager remained logged into the personal iCloud account of defendant and former Apple engineer Chang Liu after he left the company to transfer files, and later asked him for help with technical questions about Apple projects. OpenAI also accuses Apple of omitting the consequences of its own “inexplicable information-management practices” from its original complaint, noting that the company failed to properly secure its systems when employees departed, creating confusion and unwanted access issues that Apple now characterizes as theft. While this may sound a bit like a “the door was unlocked, so it wasn’t really stealing!” defense, this argument is meant to strengthen OpenAI’s argument that former Apple employees were simply trying to assist their prior colleagues. It could also help push the narrative that these weren’t truly trade secrets if they weren’t secured as such. OpenAI argues Apple hasn’t specified which “trade secrets” or confidential components were allegedly stolen, referring to them instead as “generic categories of the product-development process—such as component manufacturing, product testing, vendor and supplier relationships, and distribution channels.” OpenAI’s filing posits that Apple is using this lawsuit to slow down its competitor from innovating in AI-powered hardware, rather than focusing on its own product plans. “OpenAI has no use, need or desire for Apple’s trade secrets,” the motion argues. “OpenAI is building something entirely new and different from anything at Apple. OpenAI does have an interest in hiring the best engineers, inventors, developers and creators—many of whom have decided to leave Apple and to come to OpenAI, attracted by the innovative and exciting work the company is doing. Apple might not like that,” it reads. “…Apple should not be permitted to use a baseless and pretextual lawsuit to make up for itsshortcomings in the market for talent and retaining its employees, and its failures to integrate AI into its products,” the motion says.
Schreiber Foods Taps Ascendion to Deploy Agentic AI Across Global Business
The partnership will introduce AI agents to automate technology workflows, improve food safety monitoring and modernise ERP systems across its global network.
View
Microsoft Launches Hyderabad Cloud Region Amid AI Demand
Microsoft’s Hyderabad region has already attracted customers including Adani Group, Bajaj Finserv, HDFC Bank, and PB Pay while strengthening its AI and cloud infrastructure.
View
KDEM & GoodWorks Target ₹5,000 Crore GCC Investments, 50 New Centres in Karnataka
The collaboration also places significant emphasis on workforce development.
View
Powered by Intel, SoftBank Plays the OpenAI Long Game
Despite a $20 billion OpenAI investment in the last few months, it was Intel that delivered the biggest boost to SoftBank’s quarterly investment gains.
View
Karnataka Eyes Anthropic Partnership to Bring AI Into Public Services
The discussions included plans for an AI University, Claude certification programmes and hosting government data within Karnataka.
View
Anthropic Could Build Custom AI Chips for Claude Models, Job Listing Suggests
Anthropic recently relaunched its most capable AI model yet, Claude Mythos, which offers cybersecurity capabilities, along with Claude Fable 5. With the rising demand for AI models, the US-based AI giant could be looking to foray into a new category, as it has started hiring engineers for its Reinforcement Learning team, which will be responsible for designing custom AI chips in-house. A company spokesperson has reportedly confirmed the development as well. The AI chips are said to enhance the performance of Claude AI models. Anthropic is also reportedly in talks with Samsung about manufacturing its chips.
View
JLL Opens Hyderabad GCC, Plans to Scale to 1,600 Employees
The new facility marks JLL Business Services’ second India hub after Gurugram.
View
AI4Bharat Wants to Test Every AI Claim India Makes
“With AI systems becoming more deeply integrated into society, rigorous evaluation becomes as important as model development itself.”
View
Indian IT Stocks are Booming Again—And it Has Nothing to Do With Indian IT
Indian IT stocks are surging, but analysts warn the rally is running well ahead of the fundamentals.
View