Models, products and AI startups, near-duplicates collapsed to the most credible source
Bill Gates: AI may push some toward an 'Amish' lifestyleOn the Decoder podcast, Bill Gates called AI a “Pandora's Box” and predicted many people may choose, like the Amish, not to live at the technology frontier. He warned the next 20-year transition will bring disruption and winners and losers. Six days later, Microsoft offered buyouts to 60,000 employees.
Bridgewater CEO Nir Bar Dea discusses AI investing and marketsBridgewater CEO Nir Bar Dea discusses the US economy, rising national debt, the dollar, and how he expects AI to reshape investing, in an interview on The David Rubenstein Show. He also recounts his path from the IDF to becoming Bridgewater's CEO. The interview was recorded September 25 in New York.
Serval triples valuation to $3 billion as businesses adopt AI workflow automationReuters reports that Serval's valuation tripled to $3 billion as businesses increasingly adopt AI workflow automation. The report did not disclose the funding amount or other details of the round.
Analysis finds candidate websites full of AI-written textA New York Times analysis found that candidate campaign websites are filled with AI-written text, showing generative AI's spread into political communication.
Attacker impersonated AI bots from OpenAI, Anthropic and 30 others to steal secretsCequence researchers observed a single threat actor posing as roughly 32 AI bot identities, scanning enterprise servers for about 90 file paths including .env files, cloud credentials and SSH keys. On one October day it reached 48% of Cequence-protected organizations seeing AI bot traffic; no breach was reported.
Microsoft seeks to raise at least $25B in US investment-grade bond saleMicrosoft aims to raise at least $25 billion in its first US investment-grade bond sale in more than nine years, joining a borrowing spree funding the AI investment boom. People familiar say the final size is undecided and may change with demand.
Green AI advocate urges sustainability as top-level design concernSpeaking at GOTO Copenhagen, Charles Humble argued sustainability should be a top-level architectural concern, citing IEA data that combined electricity use by Amazon, Microsoft, Google and Meta more than doubled from 2017 to 2021 to about 72 TWh. Microsoft's 2024 report shows emissions up 30% since 2020; Google's rose 13% year on year in 2023.
Anthropic CEO Dario Amodei charts new direction for tech philanthropyCNBC reports that Anthropic CEO Dario Amodei is charting a new direction for tech philanthropy. The report provides no further details on his plans.
Kogui people of Colombia issue urgent message on artificial intelligenceThe Kogui people of the Sierra Nevada de Santa Marta in Colombia issued a statement on artificial intelligence, voiced by Mamo Vangelio Moscote Alberto and Saka Rosa Mamatacán and translated into Spanish and English. They express concern about AI development and frame it through their concept of Aluna.
US Series B funding rises but deal count fallsCrunchbase data shows roughly $48 billion has gone into US Series B-stage financings this year, a multiyear high, while round counts are below year-ago levels and heading for a multiyear low. About two-thirds of the money went to AI startups; rounds of $100 million or more made up roughly a quarter of deals, led by Prometheus at $12 billion.
Jumbo Series A rounds mask pressure on seed startupsCrunchbase reports at least 114 global Series A rounds of $100 million or more this year, over 70% to AI companies, yet US Series A rounds fell 28% year over year to 272 in Q3. Only 24% of 2023 seed companies have moved past seed, versus 55% or more through 2020, as Series A investors now expect $2M-$3M in ARR.
AI assistant startup Wabi raises $55M at $1.25B valuationAI assistant startup Wabi Inc. has raised $55 million at a $1.25 billion valuation. The funds will build out software that lets groups of people interact with AI agents to coordinate activities and make collective decisions.
Big Tech braces for AI regulation fight after US midtermsBloomberg notes that AI leaders aligned themselves with Trump, an alliance that may not hold up after next week's midterm elections, as Big Tech prepares for a fight over AI regulation.
I Used AI to Fake a Scientific Paper. Here's What HappenedAn experiment tested whether AI-generated papers can pass peer review, highlighting the flood of AI slop in academic publishing: by 2025, about 57% of articles showed LLM influence, up from 12% in 2023, and over 12% of Nature Communications reviews were fully AI-generated.
Cadence launches Maestro domain-specific model portfolio for chip designCadence unveiled Maestro, a portfolio of domain-specialized models for its engineering software and AI Super Agents, combining reinforcement learning with EDA tools. Cadence claims better design outcomes and greater token efficiency than general-purpose foundation models.
AI data centers strain Arizona's water supplyTSMC's Phoenix chip fab, seen as an economic savior, uses nearly 5 million gallons of water daily from the city supply. New hyperscale data center proposals now demand more water, power and land than Arizonans are willing to give, drawing pushback from Phoenix to Tucson.
WIRED Tests the AI Death Clock AppA WIRED reporter tried Death Clock, an app that claims to predict when you will die and help you extend your life. After a month of use, the reporter felt the experience was draining rather than helpful.
ML engineer weighs quitting job to pursue PhDA machine learning engineer with four years at a startup described work dominated by software development, DevOps and customer support rather than deep modeling, plus fatigue from agentic coding, and is considering leaving to pursue a PhD. He worries academia has worsened due to saturation and AI-generated slop, and has not decided.
Hemmingway-1.5 claims to evade AI detectorsHemmingwayAI released Hemmingway-1.5, a 27B open-weight Qwen finetune, claiming it passes GPTZero nine times in ten versus 86% for real human writing. Weights are free for non-commercial use.
Altman Says AI Is Rapidly Exposing Security FlawsAt Oracle's annual conference in Las Vegas, OpenAI CEO Sam Altman said AI systems are uncovering security vulnerabilities at a staggering pace, adding he was "shocked." Oracle also unveiled AI tools that handle complex tasks on their own, but Altman warned this capability has a troubling downside.
Claude Code reportedly injects user email into prompts, leaking it to websitesHacker News users report that Claude Code injects the user's email address into its prompt and the model then sends it to external websites, raising privacy concerns. Commenters argue Anthropic should make this configurable and that Auto Approve mode should have blocked it.
HKEX launches AI Index, Southern Asset Management authorized for ETFHKEX launched its first AI-themed index, HKEX AI, covering four segments: AI chips, infrastructure, models, and applications. The index has 40 constituents, including SMIC, Alibaba, Tencent, and next-gen AI firms like Biren Technology, Zhipu AI, and MiniMax. HKEX authorized Southern Asset Management to launch an ETF tracking the index in mainland China. Constituents must meet R&D and capex thresholds of at least 3% of sales each and sales growth of at least 5% over two years; individual weights are capped at 10% (3% for applications), with quarterly rebalancing.
AI slowdown advocate privately told investors he'd race aheadThe Washington Post reports that a figure who publicly called for an AI slowdown privately told investors he would race ahead. The report highlights a contradiction between public stance and private investment messaging.
OpenAI details how it detects frontier model misalignmentOpenAI outlined a layered monitoring system to detect when frontier models become misaligned or agents act without authorization, so responders can intervene before incidents escalate.
Developer seeks help running BiRefNet on Qualcomm chipsA developer asked the community for help deploying the BiRefNet segmentation model on Qualcomm mobile GPUs and NPUs with sub-second latency. Compilation to the NPUs keeps failing, which the poster attributes to the model's deformable convolutions, and he asked whether others have succeeded.
What Wall Street really thinks about the AI trade nowA Bloomberg podcast discusses Wall Street's current views on the AI trade, against the backdrop of market wobbles from headlines like Anthropic's S-1 and earnings from hyperscalers Oracle and Microsoft. Barclays global head of equities tactical strategies Alexander Altmann helps make sense of whether AI is immune to interest rates, historical precedents for AI capex, and the KOSPI-fication of the S&P 500.
Ropes & Gray builds transactional diligence AI with OpenAILaw firm Ropes & Gray is developing AI for transactional diligence with OpenAI; lawyers review grouped documents, summaries and reports before advising clients.
AI as Therapist: Can Chatbots Help?A DW video explores whether AI chatbots can offer genuine psychotherapy. Psychotherapist Umut Özdemir conducts an experiment to see if people can distinguish therapist responses from AI-generated ones, and experts discuss the psychological, societal and ethical challenges.
Air Liquide Q3 Revenue Grows on AI DemandAir Liquide's third-quarter revenue rose to €7.3 billion ($8.1 billion), beating analysts' €7.1 billion estimate. Its electronics unit, which supplies ultra-pure specialty gases and advanced materials to semiconductor factories, saw comparable sales grow 12%, driven by AI-related demand.
Tech Rally Stalls as Alphabet, Samsung DisappointDisappointing earnings from Alphabet and Samsung halted this week's global tech stock rally and revived doubts over whether billions of dollars in AI spending will deliver sufficient returns.
The A.I. Daily