← Back to news

T

Techmeme RSS Feed

The EU gives Google a six-month deadline to lift technical barriers to rival AI search assistants on Android and give key data to other search engine providers (Samuel Stolton/Bloomberg)

Moonshot says Kimi K2.5 builds on K2 with "pretraining over ~15T mixed visual and text tokens" and "can self-direct an agent swarm with up to 100 sub-agents" (Kimi)

Chinese startup Moonshot releases Kimi K2.5, saying the model can process text, images, and videos simultaneously and beats its open-source peers in some tests (Bloomberg)

An interview with OpenAI for Science head Kevin Weil on the team's mission, why LLMs can't come up with game-changing discoveries yet, and more (Will Douglas Heaven/MIT Technology Review)

TikTok US says it doesn't "have rules against sharing the name Epstein in direct messages" and is "investigating why some users are experiencing issues" (Bobby Allyn/NPR)

Micron plans to invest $24B over the next decade to build a 700,000-square-foot memory chip facility in Singapore, with wafer output set to start in H2 2028 (Reuters)

California Governor Gavin Newsom says he will launch a review into whether TikTok is violating state laws by censoring content critical of President Trump (Tyler Katzenberger/Politico)

The US Army signs a $5.6B, 10-year contract with Salesforce to speed up outdated practices; as part of the deal, military recruiters will get access to Slack (Wall Street Journal)

YC updated its website to remove Canada from its list of approved jurisdictions for incorporation, leaving the US, Cayman Islands, and Singapore as options (Murad Hemmadi/The Logic)

Khosla Ventures' Vinod Khosla and Ethan Choi disavow Keith Rabois' comments on an ICE shooting; Rabois said the man killed by agents was committing a "felony" (Julie Bort/TechCrunch)

A test of ChatGPT Health and Claude for Healthcare with data from Apple Health finds the chatbots provided questionable and inconsistent responses (Geoffrey A. Fowler/Washington Post)

Meta says it will test new paid subscriptions that give people access to exclusive features on its apps and plans to scale Manus as part of those offerings (Aisha Malik/TechCrunch)

StepFun, a Chinese AI startup that develops LLMs and has partnered with automaker Geely and smartphone brands like Oppo and Honor, raised a ~$717M Series B+ (Eudora Wang/DealStreetAsia)

Tim Cook attending a private White House screening of the documentary Melania hours after US agents killed a man suggests horrible judgment, or worse, cowardice (M.G. Siegler/Spyglass)

Billionaire Tom Steyer, who is running for California governor, says he plans to vote for the one-time 5% tax on billionaires if it reaches the November ballot (Eliyahu Kamisher/Bloomberg)

A group of YouTubers with a combined 6.2M subscribers adds Snap to a class action lawsuit, alleging the company trained its AI systems on their video content (Sarah Perez/TechCrunch)

Google agrees to pay $68M to settle a class action lawsuit claiming that Google Assistant inappropriately recorded private conversations on users' devices (Jonathan Stempel/Reuters)

Wikimedia data: iOS 26 adoption was ~50% in January, below iOS 18's 72% in 2025, as Apple slows auto-updates; Statcounter had 15% after missing Safari changes (John Gruber/Daring Fireball)

Zoom's shares closed up 11.28% after analysts at Baird estimated that Zoom's 2023 investment in Anthropic could be worth $2B to $4B, depending on dilution (Nate Wolf/Barron's Online)

Zocks, which offers an AI assistant for financial advisers, raised a $45M Series B co-led by Lightspeed and QED Investors, taking its total funding to $65M (Mary Ann Azevedo/Crunchbase News)

Anthropic rolls out a new extension to MCP to let users interact with apps directly inside the Claude chatbot, with support for Asana, Figma, Slack, and others (Robert Hart/The Verge)

UpScrolled, a platform for sharing photos, videos, and text that claims political impartiality, says user growth has surged since TikTok's US takeover (Emma Roth/The Verge)

Qwen releases Qwen3-Max-Thinking, its flagship reasoning model that it says demonstrates performance comparable to models such as GPT-5.2 Thinking and Opus 4.5 (Qwen)

In a 38-page essay, Dario Amodei warns of civilization-level damage from superintelligent AI, questioning whether humanity has the maturity to handle such power (Axios)

Upwind Security, which filters out false security alerts in cloud apps to let security teams address actual risks, raised a $250M Series B at a ~$1.5B valuation (Angus Loten/Wall Street Journal)

TikTok US says it's working to restore services after a power outage at a data center and that any algorithm changes users noticed were likely due to the outage (Dan Whateley/Business Insider)

Microsoft unveils the Maia 200, its 2nd-generation AI accelerator built on TSMC's 3nm process, deploying today in its Azure US Central data center region (Tom Warren/The Verge)

Sources: the US DOT plans to use Gemini to draft federal regulations, cutting the process to just 30 days; the DOT used it to draft a still-unpublished FAA rule (Jesse Coburn/ProPublica)

The EU Commission says WhatsApp's Channels fall under the DSA's Very Large Online Platform rules, joining Instagram and Facebook, requiring it to address risks (Anupriya Datta/Euractiv)

Source: OpenAI targets ~$60 per 1,000 views for ChatGPT ads, on par with live NFL broadcasts and above Meta's sub-$20 CPM, while offering little conversion data (The Information)

Nvidia announces its Earth-2 Medium Range weather model, built on its Atlas architecture, claiming it outperforms Google DeepMind's GenCast in 70+ variables (Tim De Chant/TechCrunch)

Apple unveils a second-gen AirTag with an updated speaker and the same second-gen Ultra Wideband chip as in the iPhone 17, costing $29 for one and $99 for four (Ryan Christoffel/9to5Mac)

Intel Panther Lake laptop CPU review: the 18A-based Core Ultra X9 388H outperforms Apple's M5 and AMD's top-tier Strix Halo in some tests like 4K video exports (Antonio G. Di Benedetto/The Verge)

As tech CEOs stay silent after two fatal shootings, 450+ workers at Google, Meta, OpenAI, Amazon, and others urge CEOs to cancel any ICE contracts and speak out (Axios)

How the explicit deepfake ecosystem evolved from images to high-quality video; an analysis shows 1.4M+ Telegram accounts signed up to 39 deepfake creation bots (Matt Burgess/Wired)

Nvidia invests an additional $2B in CoreWeave to help speed up adding 5GW+ of AI computing capacity by 2030 and deploy Nvidia's new Vera CPU; CRWV jumps 9%+ (Bloomberg)

TSMC's conservative capex approach is resulting in supply-demand imbalances and foregone revenue, and hyperscalers must build up Samsung or Intel as its rivals (Ben Thompson/Stratechery)

Sources: NY-based Tandem, which uses AI to automate the paperwork for writing and receiving medical prescriptions, raised a $100M Series B at a $1B valuation (Rebecca Torrence/Bloomberg)

Interviews with 100+ therapists and psychiatrists on clients' AI chatbot usage show, while there are some upsides, conversations also deepened negative feelings (New York Times)

OpenAI says nearly 1.3M weekly users are discussing "advanced topics in hard science" in January, with an average of 8.4M ChatGPT messages on those topics (Ashley Gold/Axios)

Ricursive, founded by ex-Google researchers to automate advanced chip design, raised $335M from Sequoia, Radical, Lightspeed, and others at a $4B valuation (Cade Metz/New York Times)

Morgan Stanley: UK companies reported that AI led to 8% net job losses over the past 12 months, above Japan's 7%, Germany's 4%, Australia's 2%; the US added 2% (Irina Anghel/Bloomberg)

Documents: China's military is focused on harnessing AI to deploy swarms of drones, robot dogs, and other autonomous systems, learning from hawks and coyotes (Josh Chin/Wall Street Journal)

The EU opens a formal DSA investigation into xAI over Grok generating sexualized images of women and children; xAI faces fines of up to 6% of global revenue (Barbara Moens/Financial Times)

Claude Code can feel daunting, and most people's problems are not software-shaped, but it is clearly autonomous and the home-cooked app renaissance is great (Jasmine Sun/@jasmine's substack)

Digital Realty, QTS, and NTT Data warn the data center industry is doing a poor job of combating local opposition; 24+ US projects were blocked in January alone (Rafe Rosner-Uddin/Financial Times)

Chinese autonomous trucking company DeepWay raised ~$173M ahead of its planned Hong Kong IPO, filed in November 2025; DeepWay reported ~$215M in H1 2025 revenue (Dong Cao/Bloomberg)

Quantum computing company IonQ acquires US chipmaker SkyWater for ~$1.8B, paying $35/share, in IonQ's biggest deal yet; SkyWater will operate as a subsidiary (Lauren Thomas/Wall Street Journal)

Sources: SoftBank halts talks about a ~$50B acquisition of US data center operator Switch, a setback to Masayoshi Son's goal to roll out Stargate infrastructure (Bloomberg)

London-based enterprise AI video startup Synthesia raised a $200M Series E led by GV at a $4B valuation, up from $2.1B after raising $180M in January 2025 (Nat Rubio-Licht/The Deep View)