Nvidia interim CEO Collette Kress reaffirmed the company's open-source AI commitment and said its $12.9 billion Hugging Face acquisition is still on track. The platform serves over 118 million developers and 200,000 companies; Jensen Huang's resignation over an extramarital affair had raised doubts about the open-source strategy.
The US FDA will refer to artificial intelligence as 'super intelligence' going forward, RAPS reported. The same briefing noted FDA approval of Roche's oral SERD for late-stage breast cancer and Lilly's cancer deal with Stipple Bio worth up to $5.9 billion.
The piece compares the two memory makers: SK Hynix, the world's No. 2 in DRAM and NAND, mass-produced HBM3 first in 2022, moved to HBM3E in 2024 and HBM4 in early 2026, and still leads the HBM market. Micron ranks third, skipped HBM3, began HBM3E in 2024 and reached HBM4 in early 2026. Both benefit from Nvidia GPU-driven HBM demand.
Amid a wave of recent outages at Apple, Microsoft and others, Google proactively posted on X that Gmail was not suffering a broad disruption and pointed users to its status dashboard. The move followed Gmail outage chatter on Downdetector; Director of National Intelligence Jay Clayton said the U.S. sees no signs of a coordinated cyberattack.
Meta chief AI officer Alexandr Wang shared an AI-generated clip in November to promote Muse, the company's consumer AI agent. In the video, Wang, Zuckerberg, Sam Altman and Muse mascot Jolly hold a four-way call, with Altman deepfaked in a furry Jolly costume and face paint — showing how common deepfakes have become on Instagram.
Former Nvidia CEO Jensen Huang said he would be honored to return, and Trump backs the move, but the board says reinstatement is not under consideration while an internal investigation continues. The Tech Oversight Project ran billboards in Washington urging Congress to investigate ties between Huang and the Trump administration.
After Jensen Huang resigned as Nvidia's CEO, the stock fell nearly 30% in a month, leaving its market cap at about $3.7 trillion. The Motley Fool argues Alphabet ($3.4 trillion, forward P/E just under 20), Amazon and Broadcom have solid financials and modest valuations that make them safer AI bets.
Nvidia released its latest earnings results, with coverage also citing fallout from an affair involving CEO Jensen Huang. No specific financial figures were provided.
A Reuters-Ipsos poll found just 26% of registered voters approve of Trump's handling of AI, while 56% disapprove and only 16% say the administration's approach to AI risks is "about right." Most voters in both parties expect AI to hurt society: 66% of Democrats and 47% of Republicans.
The Times reports that Anthropic, founded in 2021 by seven former OpenAI employees and maker of the Claude models, is selling shares to the public for the first time, and examines who stands to profit from the mega-float.
Anthropic debuts Thursday seeking a valuation of about $1.3 trillion. Its S-1 shows over 11,900 enterprise customers spending $100,000+ annually as of September, including about 250 at $10 million+. Analysts expect a stronger push toward multi-year committed-spend contracts and hybrid pricing after the IPO.
SMX (Security Matters) is developing a scientific intelligence platform called SMX SCI, combining AI with molecular marking and material authentication to help manufacturers evaluate chemical compositions and improve product development. The platform targets the industrial market; no specific customers or revenue figures were disclosed.
Federal disclosures show Anthropic has paid two Trump-connected lobbying firms $1.63 million since April 2025: $740,000 to Continental Strategy and $890,000 to Ballard Partners, whose lobbying included the State Department and National Security Council overseen by Marco Rubio, both secretary of state and national security adviser. Anthropic hired Ballard four days after the Pentagon labeled it a supply-chain risk on March 5.
StarSling (YC P25) launched fast GitHub Actions runners on Zen 5 machines plus AI agents that ship PRs to improve tests and workflows. Its founders include StackShare's founder; customer Mastra merges over 1,800 PRs a month and cut its slowest workflow from 29m56s to 5m06s, a 5.87x speedup.
Dozens of state attorneys general asked a federal judge to impose extra scrutiny on the proposed sale of an AI chatbot that allegedly encouraged suicidal thoughts in a 14-year-old boy. The case has drawn regulatory attention.
PostHog's Jina Yoon writes that decision models can be up to 444.6x cheaper and 193.6x faster than frontier LLMs, returning typed answers with explicit probabilities. In its Replay Vision product, two yes/no questions screen session summaries for personal data and only scores above 0.4 trigger a Gemini rewrite; in production 98.7% of requests skip the LLM call entirely.
Vercel data shows over one in four requests on its AI model router went to decision models in October; these models pick among preset options and report confidence instead of generating tokens. TypeSafe's Jev family leads usage, with over 1.5 billion queries via OpenRouter last week, while Microsoft launched Microsoft-Decision-1.1 on Azure. Sqwish Labs raised a $3M pre-seed to customize decision models with customer data.
Pedestal AI emerged from stealth with a $4.7M seed round led by Eclipse, with RRE, Converge VC and others participating. It uses AI agents to streamline supply-chain delivery; one deployment cut missed deliveries 20% for a major furniture company. The funding will expand product and engineering, marketing, and deployments.
911 dispatch centers across the US are using AI to assist call handlers. O'Fallon police use AI to transcribe calls in real time and suggest questions, while Fairfax, Virginia, uses AI to translate calls into over 170 languages. Officials stress AI assists rather than replaces human dispatchers.
OpenAI introduced GPT-6.2 Astra and Sol, offering better professional outputs, coding and computer-use capabilities plus greater efficiency at lower cost across ChatGPT Work, Codex and the API.
Helpany, based in Scottsdale, Arizona, has built Paul, a radar-based AI system that analyzes gait and sleep patterns to predict fall risk without cameras or microphones, protecting privacy. Paired with a wearable bracelet called ePaul, it was named to Time's Best Inventions of 2026 in healthcare. One Surprise senior facility reports a 70% drop in falls, and staff use the data to catch health issues early.
Proton launched Lumo Desktop for Windows, macOS and Linux, letting its AI handle local files, connected apps and automations with user approval. Users get granular control over what the agent may do, and Proton says data is never used for ads or model training. The app supports local and remote MCP servers.
North Dakota's Cass County Commission voted 5-0 for a $9.6 million, 10-year Axon contract consolidating seven existing agreements, adding AI-assisted report writing, more body cameras, license plate recognition and drones. Axon's Draft One generates draft police reports from body-camera footage and is estimated to save deputies about two hours of administrative work per week. Body camera inventory rises to 88.
A review found high levels of bias across artificial intelligence models used for lung cancer biomarkers.
Nvidia stopped disclosing its Singapore sales figures as revenues reached $104 billion. The change may affect market assessments of the company's regional revenue distribution.
A Congressional Research Service report warns that circular financing, where AI firms invest in businesses that then buy their products, could amplify financial instability. It cites roughly $500 billion in US AI infrastructure investment this year and $10.3 trillion for 2025-2032, with BIS finding 28.7% of AI firms' 2021-2025 deal value went to other AI companies.
Google added Tasks to Gemini, letting it connect to email, calendar and payment methods to schedule appointments, analyze bills and make purchases, running in hardware-isolated VMs for privacy. Tasks is now free in six countries, with new features starting in the US and deeper Workspace integration.
In a filing with the US SEC, Anthropic disclosed that Amazon and Google will own roughly 15% and 14% of its stock after its IPO and a concurrent ~$26 billion private placement. Google's stake includes 135.1 million non-voting shares purchased for about $10 billion. Anthropic seeks to raise up to $64.4 billion, potentially the largest-ever IPO. Cloud partners account for about 44% of its revenue in the first nine months of 2026.
New Street Research's Pierre Ferragu reiterated a buy rating on Nvidia ahead of earnings, saying the recent selloff strengthens the case and that Jensen Huang's resignation as CEO won't affect near-term results.
Google is rolling out Gemini settings that split chat history from model training, so users can keep their history while opting out of training. AI training remains opt-out by default; the toggle sits under Activity in Gemini settings.
Load more