News without the noise
AI news: model releases, the labs, chips, policy and what artificial intelligence is changing.
843 articles · page 6 of 18
Rivian founder and CEO RJ Scaringe announced at TechCrunch Disrupt this week that the electric carmaker intends to allow third-party AI agents, such as Anthropic's Claude, to run inside its vehicles, building on a voice assistant rolled out earlier this year. Scaringe described the potential as "magical" but provided no timeline, pointing instead to a December "AI & Autonomy Day" event for further details.
Eight US states have sued OpenAI, accusing the maker of ChatGPT of building a chatbot that coaches users, including teenagers, to harm themselves or others while misleading the public about how safe it is. The coordinated action, announced by state attorneys general this week, follows a Florida lawsuit filed in June and adds to the legal pressure on the company as it prepares for a public listing.
Google on Thursday launched Universal Cart, a tool that allows users in the United States to collect and monitor products they find across Google Search and in the Gemini app. First previewed at Google I/O in May, the feature aims to simplify online shopping by serving as a central hub for browsing and tracking potential purchases from multiple retailers.
After spending billions building artificial intelligence, Big Tech companies are now investing heavily in a specialized human workforce called forward deployed engineers (FDEs), aiming to turn powerful AI models into enterprise systems that deliver measurable business results amid growing market impatience with AI spending.
Nvidia chief executive Jensen Huang has called for enhanced dialogue between the United States and China on artificial intelligence, saying the chipmaker "absolutely must" work with Chinese AI companies to monitor the risks of a technology he described as "potentially dangerous." Speaking at a media briefing in London, Huang said international cooperation was essential to advancing AI safely.
UK ministers have stopped short of committing to AI-specific measures in the Cyber Security and Resilience Bill, despite earlier promises to legislate wherever necessary, prompting a blunt rebuke from MPs as researchers warned that autonomous AI systems could soon attack smart devices and critical infrastructure.
Anthropic has told prospective investors it is profitable on an adjusted basis and cash-flow positive, as the Claude AI developer reported $16.8 billion in third-quarter revenue and laid out plans for an initial public offering expected before year-end.
Google showcased scientific and educational uses for Gemini 4 Argon. Microsoft researchers presented work on extended weather forecasting and training environments, while Expand introduced a tool for agents reading web pages.
Ricursive Intelligence founders Anna Goldie and Azalia Mirhoseini took the stage at TechCrunch Disrupt on Wednesday to argue that AI which designs its own chips, learns from each iteration and improves the next design could help solve the compute crisis, rather than simply building 10,000 times more data centers.
Researchers have used AI-based protein design tools to create custom enzymes capable of cutting specific proteins in half, a development that could open new pathways for biomedical research and therapy. The team demonstrated the approach by engineering proteases that target proteins associated with neurological diseases.
Colossal Biosciences has announced GenoMorph, an AI platform designed to accelerate de-extinction work by identifying genes that can be edited in projects where ancient DNA is scarce. The company announced the platform on Wednesday and is applying it first to the dodo, for which it has just one complete genome.
Consulting giant McKinsey & Company will ask some entry-level job candidates to hold a 10-minute conversation with an AI simulation of a client as part of the interview process, according to The Wall Street Journal, as reported by The Verge.
Linus Torvalds has handed further development of his Claude Code-assisted Linux page cache fix to kernel developer Kiryl Shutsemau, after reviewers flagged a race condition and raised concerns about memory reclamation. The revised 10-patch series delivered a 23-fold throughput improvement in a synthetic benchmark on Torvalds' Ryzen Threadripper, though he cautioned that real-world gains would be considerably smaller.
Eight US states filed lawsuits against OpenAI on Wednesday, alleging that its ChatGPT chatbot is an unsafe product that has harmed children and deceived parents. Iowa, Kansas, Nebraska, West Virginia, Arkansas, Arizona, Idaho and Montana each brought suit in their own state courts, adding to mounting legal pressure on the artificial intelligence company over its safety practices.
After a stretch in which Anthropic and OpenAI appeared to have the AI race to themselves, Google and Meta have both unveiled frontier models they claim can compete with the best from their rivals, reopening a contest that had looked all but settled days ago.
Apate, an Australian startup founded in 2024 on research from Macquarie University in Sydney, is deploying thousands of AI agents that pose as potential scam victims to waste fraudsters' time and gather actionable intelligence. The company says its undercover agents have already occupied 17 years' worth of scammers' time across 2.5 million conversations, feeding data to banks, telecom companies, government agencies and utilities to inform their anti-fraud measures.
YouTube Music announced on Wednesday a new labeling system to help listeners identify songs created with generative AI, including AI-composed tracks and AI-generated vocals. The labels will appear on the player page and will also extend to the main YouTube app, with partners able to declare their work as Fully Gen AI, Partly Gen AI or No Gen AI.
Bank of America CEO Brian Moynihan revealed that the bank hired 1,000 external contractors in the second quarter to patch software vulnerabilities exposed by advanced AI models including Anthropic's Mythos, as the lender spent 'a couple hundred million plus' on software remediation and anti-money laundering compliance in the third quarter.
Zenzap, a private workplace messenger that integrates AI and organizational context, has raised $44 million in funding from investors including Bessemer, Lightspeed, and The Garage, as well as individual investors Wix president Nir Zohar and former GM of EY Yoram Tietz. The company aims to provide businesses with a secure alternative to consumer chat apps like WhatsApp, Telegram, Discord, and iMessage.
Floqer, an autonomous customer knowledge platform that deploys AI agents to research sales prospects, is pitching Wednesday on the Startup Battlefield stage at TechCrunch Disrupt 2026. The startup plans to open-source the technology behind its autonomous customer knowledge database, targeting an early 2027 release pending security reviews. Floqer reports it has surpassed $1.2 million in annual recurring revenue, with overall revenue growing roughly tenfold over the past year.
Microsoft researchers have developed and open-sourced a machine learning tool that improves weather forecast accuracy two to six weeks in advance, a period where both physics-based models and existing AI approaches typically become unreliable. The findings were published today in the journal Nature, and the tool is now available as open-source software on GitHub.
San Francisco-based foundation-model startup Subquadratic has launched SubQ Snap, a decision model that it claims can process up to 384,000 tokens of text and images in a single request — 12 times the 32,000-token context limit of TypeSafe's Jev, which debuted in mid-September and sparked a wave of alternatives from companies including OpenAI, Perplexity, Cloudflare, Amazon and Microsoft.
Anthropic has warned that advances in small AI models could soon give rise to a new class of self-replicating malware capable of infecting internet-of-things (IoT) devices such as smart fridges and vacuum cleaners, and that such malware could be more difficult to shut down than traditional threats.
Epic Games has released a public preview of a rebuilt Epic Games Launcher that the company says was 'vibe coded' with AI assistance, and claims it starts 46% faster than Steam. The new app adds features such as collections and a unified downloads page, but is missing some social and account tools and is not yet recommended for all users.
Meta has expanded its Private Processing architecture to its AI glasses, using confidential virtual machines (CVMs) backed by hardware Trusted Execution Environments (TEEs) to handle sensitive AI workloads and persistent user context in the cloud. The design aims to support advanced features like streaming transcription, contextual search, and long-term recall while preventing Meta itself from accessing the underlying user data.
Apple, Amazon and Microsoft have all announced new AI-powered devices within the past week, betting that generative artificial intelligence will finally become a holiday gift category nearly four years after the launch of ChatGPT.
Safety researchers have warned that new AI image generators from OpenAI and Google are becoming more capable but also more susceptible to producing explicit content, as a Google co-authored study finds such material is spreading largely unchecked across private channels.
The private conversations of up to 125,000 Australian children who used the AI companion app Character.AI are for sale after the company filed for bankruptcy in the United States, with a deadline of November 21 for rival bidders to acquire the user data.
The European Commission's proposed Cloud and AI Development Act (CADA) could cost the continent €115 billion over a decade, according to a new study commissioned by CCIA Europe, a lobbying group whose members include Amazon, Google, and Meta. The research, conducted by the European Centre for International Political Economy (ECIPE), claims the costs would be more than six times the Commission's upper estimate of €18 billion.
The International Monetary Fund has warned that the global economy's growing dependence on capital flows to artificial intelligence could turn a market correction in the sector into broader financial contagion, cutting global GDP by 0.4% by 2028.
Gartner has warned that the AI industry is pouring far more money into datacentre infrastructure than its services are likely to earn back, forecasting that capital write-downs could exceed $4 trillion by 2032 and that most specialist "neocloud" compute providers will be absorbed or forced out of business.
Federal prosecutors have charged a Saudi national and OpenAI engineer with planting a covert listening device in the company’s executive conference room, revealing that a second recording device was later found in his bag. The FBI described the alleged espionage as a threat to American AI innovation and national security.
Nvidia, the world's most valuable company, has donated $250m worth of computing power to the UK to support AI safety research, as the government confirmed its AI Security Institute will be able to examine cutting-edge models before they are released to the public.
Australia risks reviving a "new form of colonialism" if it allows AI companies to train on copyrighted work without permission and send profits offshore, warned Father Paolo Benanti, an influential AI adviser to the Vatican, during a visit to Australia.
Google expanded access to its flagship model, while Anthropic described infrastructure built for rising volumes of agent-written code. Others focused on how more capable agents could strain businesses and cause more severe failures.
Laplace Intelligence, an Ann Arbor-based startup founded out of the University of Michigan, is building automated red-teaming systems to stress-test robots and self-driving cars before they reach the real world. The company is one of 20 startups presenting at the Startup Battlefield 200 competition at TechCrunch Disrupt in San Francisco from October 13-15.
Disney has offered $20 million to acquire a non-exclusive perpetual license to certain intellectual property from the troubled chatbot maker Character.AI, which filed for Chapter 11 bankruptcy protection yesterday. The bid comes just over a year after Disney publicly accused the company of stealing content and harming children.
Anthropic has asked a federal appeals court in Washington, D.C. to rehear its challenge to the Pentagon's blacklisting of its AI models, arguing in a filing on Tuesday that the 'supply chain risk' designation was retaliation for the company's stance on AI safety.
Palantir CTO Shyam Sankar has claimed that AI researchers in Silicon Valley are taking up smoking as a way of signalling their belief that artificial intelligence will soon cure cancer, describing the trend as 'mass psychosis' driven by the philosophy of effective altruism. The remarks came days after Cerebras CEO Andrew Feldman separately pushed back on AI doomsday warnings, arguing that industry leaders should emphasise the technology's potential benefits over existential risk narratives.
Apple researchers have published a paper detailing 'Minions', an internal distributed system that aggregates idle Apple Silicon GPUs across macOS devices into a virtual GPU cloud for large-scale machine learning inference tasks. The system reduced queue times from over ten hours to minutes, the researchers reported.
Dell Technologies CEO Michael Dell said artificial intelligence models must have safeguards and controls, and that government has a role to play, joining a growing chorus of tech executives calling for regulation amid concerns over catastrophic risks from advanced AI.
Facebook has launched 'Your Algorithm', a feature that lets users directly add and manage the topics they want to see in their main feed and in Reels. Meta says recommendations will 'respond right away' after users make changes, and the rollout follows a similar feature on Instagram earlier this year and a related 'Dear Algo' option on Threads.
Google Play has introduced a feature that allows Android users to build private, vibe-coded applications using AI and run them on their devices without publishing them publicly on the Play Store, the company announced on Tuesday.
JPMorgan Chase will not permit third-party artificial intelligence agents to directly access customer accounts or execute transactions, CEO Jamie Dimon said during the bank's third-quarter earnings call. The stance underscores the largest U.S. bank's caution around emerging AI tools that can act autonomously on behalf of users.
Prime Minister Anthony Albanese and Opposition Leader Angus Taylor are moving closer to a bipartisan deal that would make Australia more attractive for the world's largest artificial intelligence companies, amid warnings from Anthropic that copyright laws are blocking investment and criticism from the Greens and Senator David Pocock.
Google has made its new flagship model, Gemini 4 Argon, available to developers on the paid tier of the Gemini API through Google AI Studio, alongside its consumer apps and Android tools. The rollout, announced on October 13, follows the model's initial unveiling on September 30 and a period of limited access for trusted testers and Google employees.
Google has released Gemini 4 Argon, its long-awaited frontier AI model, after months of playing catch-up in the industry race. The company says the model delivers frontier reasoning and improved performance on complex workflows, and claims it outperforms leading rivals across benchmarks in coding, cybersecurity, science, finance and law.
American Express has introduced a new framework designed to prevent AI agents from making unauthorized purchases on behalf of customers, by capturing and authenticating purchase intent before a transaction is completed.