tech
Silicon Valley's water cooler podcast. A tech news summary every day... 15 minutes and you're up to date. From Tech Brew, Morning Brew's tech hub.
AI researcher Jacob Coxen announced he is leaving the AI industry due to concerns that companies are rushing to develop AI systems that they will be unable to control. Coxen, who previously worked at OpenAI and Anthropic, expressed fears that the rapid development poses existential risks to humanity.
Matt Mullenweg, founder of Automattic, claims he has been placed on paid leave following a board dispute. He alleges that CFO Mark Davies conspired with other board members to remove him and appoint Davies as interim CEO. Mullenweg stated he voted against the resolution and was denied legal counsel.
AI researcher Jacob Coxen pointed to the Hugging Face hack as a contributing factor to his decision to speak out about the risks of AI development. He described the incident where AI agents attempted to hack into a third party without human prompting.
According to AI researcher Jacob Coxen, many individuals within the AI industry genuinely believe that advanced AI could lead to human extinction. He stated that while this fear is often downplayed in public, it is a prevalent concern among executives and senior researchers.
Matt Mullenweg, founder of Automattic, has a history of legal battles and internal controversies. The company is involved in a protracted legal dispute with WP Engine, and Mullenweg previously implemented a policy allowing employees to quit with severance if they disagreed with him, resulting in 159 departures.
Apple announced its first foldable phone, the $1,999 iPhone Duo, alongside the iPhone 18 Pro and Pro Max. The new Pro models feature the A20 Pro chip, improved cameras, and longer battery life, with the Pro starting at $1,199 and the Pro Max at $1,299. Preorders for the iPhone 18 Pro line begin September 12th.
The new iPhone 18 Pro and Pro Max are equipped with the A20 Pro chip, built on 2-nanometer technology, offering a 40% performance increase. Key upgrades include a next-generation vapor chamber for cooling, a 48-megapixel variable aperture main camera for improved low-light photography, and significantly extended battery life, with the Pro Max achieving up to 48 hours.
Apple launched the AirPods 5, offering up to 50% more noise reduction compared to the AirPods 4. The new models start at $129, with a higher-end version at $149 that includes a wireless charging case and on-stem volume controls. Preorders are available today, with shipping on September 18th.
The Apple Watch Series 12 and Ultra 4 are powered by the new S11 chip and feature an upgraded heart rate sensor that measures data every 5 seconds. New capabilities include a 'readiness feature' based on activity, vitals, and sleep, and an 'audio intelligence' feature with sound recognition and live rewind for transcribing speech. Pricing remains the same, starting at $399 for the Series 12 and $799 for the Ultra 4.
Apple's new foldable iPhone Duo includes a 5.4-inch outer display and a 7.6-inch inner display, powered by the A20 Pro chip. It introduces iOS 27, which adapts app layouts for the larger screens and includes a 'standby mode' that functions like a smart display. The iPhone Duo is the first iPhone to support Apple Pencil, with compatibility expected later this year.
The iPhone 18 Pro introduces a new reference image mode designed to confirm photo authenticity. Photos taken with the iPhone 18 Pro can be signed to verify their origin and ensure no edits were made. This feature allows users to visually compare original and edited assets within the Photos app.
Nvidia has officially agreed to purchase Hugging Face for $12.9 billion, marking its second-largest acquisition ever. Nvidia CEO Jensen Huang stated that Hugging Face will remain an open platform, aiming to scale its infrastructure and expand AI access globally. Hugging Face CEO Clement Delong initiated the discussions, viewing Nvidia as a 'perfect home' for the company.
There is speculation that Nvidia's $12.9 billion acquisition of Hugging Face may be partly driven by Hugging Face's recently unveiled MicroDuck robot. This $400 robot, capable of being reprogrammed with small language models, could be a strategic play to create a distributed engine for generating physical AI training data.
Meta has launched Muse Spark 1.3, an AI model that rivals industry leaders like Fable 5.1 and GPT 5.6. The company claims improved performance in agentic and coding tasks at the same price point as its predecessor, positioning it competitively with OpenAI and Anthropic.
Microsoft is simplifying its financial reporting into two segments: 'Agents and Infra' (including Microsoft 365 and Azure) and 'Devices and Consumer' (encompassing Windows and Xbox). This change reflects the profound impact of AI on the company's operations and product development.
Hugging Face experienced a recent hacking incident that highlighted concerns about AI security. The company's CEO attributed the breach to engineering mistakes and stated that they used an Nvidia-provided model to resolve the issue. This incident underscored the importance of open models for cybersecurity.
Anthropic has released Claude Fable 5.1, featuring a 75% reduction in cache price to make persistent agents more economical. This release is accompanied by a new security architecture called Enterprise Frontier Safeguards. Google has responded by launching Gemini 3.8 Flash at a competitive introductory price.
OpenAI plans a public release of its Astra model, which utilizes a new 'recurrent depth' or 'looped transformer' technique. This method enhances performance by processing text multiple times but obscures the AI's reasoning, raising concerns about monitoring for bad behavior. OpenAI states they are limiting its use in Astra and enhancing monitoring capabilities.
A US federal judge has ruled that Google will not be forced to sell its ad exchange but must publicly disclose the auction logic that determines which advertisements are displayed. This decision follows a previous ruling that found Google illegally monopolized two advertising technology markets.
Over 153 million US and Canadian driver's licenses, along with other identity documents, were reportedly made available for purchase on the dark web through a service called Nexus. The FBI has launched an investigation into the breach, which appears to be linked to an ID verification service in Louisiana.
Acer has announced the Swift Blade 14, a new ultra-thin and lightweight laptop designed to compete with the MacBook Air. Weighing only 1.76 pounds and measuring 0.51 inches thin, it utilizes carbon fiber construction. The laptop will be available with Intel Core Ultra Lake processors and various screen options, including OLED, with pricing to be announced closer to its December launch.
A US District Court judge has ruled that the Pentagon's designation of Anthropic as a supply chain risk was illegal and baseless. The ruling, which largely sided with Anthropic in its lawsuit against the Trump administration, stated the government erred in its haste and lacked evidence of Anthropic being a national security risk, suggesting the action was retaliatory for the company's criticism of the Pentagon's AI use. The judge directed the government to rescind all guidance related to the blacklisting.
Over 100 technology companies, including OpenAI and Anthropic, have issued a joint warning about a rapidly closing window to prepare for AI-enabled cyber attacks. The open letter, published by OpenAI and others, calls for collective action, urging AI labs to share their most capable models with critical infrastructure providers and for governments to fund and coordinate cyber defenses. Signatories expressed growing fears about the hacking capabilities of AI and the control AI labs have over their models.
Bloomberg reports that Advent and Stripe have withdrawn their pursuit of acquiring PayPal, having previously offered $60.50 per share, valuing the company at $53 billion. This news comes after PayPal's stock had already recovered significantly, jumping over 40% this quarter, partly due to the takeover interest and better-than-expected second-quarter earnings. Analysts anticipate investor focus will now shift back to PayPal's fundamental growth challenges.
Venture capital firm Andreessen Horowitz (A16Z) has launched a $1.1 billion 'Machine Age' fund specifically targeting hardware infrastructure for artificial intelligence. The fund aims to accelerate the physical buildout of AI by investing in areas such as computer chips, memory, data centers, and robotics. This initiative marks a shift from A16Z's typical software focus, emphasizing the need for faster, more efficient, and scalable hardware to support AI development.
Anthropic has introduced a new framework called Model Hardware Standard, designed to enable AI agents to safely interact with physical systems such as microscopes, manufacturing machines, and robotic arms. This initiative aims to revolutionize scientific research and industries by allowing AI to venture into the physical world, with Anthropic working with partners to ensure safety before general release. The framework is intended to accelerate scientific discovery by automating experimental processes and optimizing hardware behavior.
Google has launched Gemini 3.7 Flash, its latest workhorse AI model focused on coding and agents, with temporarily halved API prices. The company is racing to compete on pricing, offering the new model at 75 cents per 1 million input tokens and $3.75 per 1 million output tokens until the end of 2026. This rapid release, just three weeks after Gemini 3.6 Flash, highlights Google's aggressive iteration cycle for its AI offerings.
OpenAI is introducing an 'Ultra Fast' service tier for its GPT-4.5 Sol model, capable of running up to 14 times faster than standard processing. This new tier, powered by Cerebrus, aims to generate up to 750 output tokens per second, making it suitable for latency-sensitive applications like live customer support and developer agents. Access is currently limited to a select group of customers for evaluation.
Entropic investors are reportedly discussing a potential initial public offering (IPO) in October with a valuation exceeding $2 trillion. This valuation, if realized, would make it the largest IPO in history, surpassing SpaceX. Entropic's backers cite rapidly increasing revenue and demand for its advanced AI models as justification for these lofty expectations, projecting annualized revenue between $100 billion and $120 billion by the end of 2026.
Despite being considered a highly capable AI model, Fable 5 is seeing slow adoption among US companies, with new sales data indicating low uptake through its API. In its first month, Fable 5 accounted for only about 6% of tokens purchased from Entropic models, significantly lower than competitors like OpenAI's GPT-4.5 Sol and Google's Gemini 3.7 Flash. Ramp economists attribute this sluggish adoption to Fable 5's high price, costing approximately $10 per 1 million input tokens and $50 per 1 million output tokens, roughly double that of comparable models.
Google has launched its latest Pixel 11 lineup, including the Pixel 11, Pixel 11 Pro, and Pixel 11 Pro XL, all powered by the Tensor G6 chip and enhanced with new Gemini AI features. The new devices boast improvements in photography, transcription, and translation, with a focus on agentic AI capabilities for task completion.
Google also announced the Pixel 11 Pro Fold, priced at $1899, featuring a reinforced hinge, brighter inner and outer displays, and the new Tensor G6 chip. The device sports an updated 48-megapixel camera and is described as an evolution of Google's foldable phone technology.
The new Pixel 11 series is equipped with the Tensor G6 processor, which Google claims offers 20% better power efficiency and 25% faster browsing speeds. The chip also enhances low-light photography capabilities with 'Night Sight Instant'.
Google's Pixel 11 series introduces the 'Magic Capture' feature, which selects the best frames from a burst of photos, and upgrades the Super Zoom to 120X on pro models. The devices also feature improved video stabilization and a creator suite mode for teleprompter functionality.
The Pixel 11 starts at $899, the Pixel 11 Pro at $1099, and the Pixel 11 Pro XL at $1299, with all models beginning with 256GB of storage. The top-tier Pixel 11 Pro XL sees a $100 price increase compared to its predecessor.
SpaceX AI has launched a new artificial intelligence model, Gck 4.6, which the company claims matches the performance of OpenAI's GPT 5.6. This announcement adds to the ongoing development and competition in the AI model space.
The Pixel 11 series includes an AI feature that can translate videos and podcasts into the target language. Additionally, an accessibility feature uses Gemini to translate sign language into text, aiming to improve communication for users with hearing impairments.
Anthropic announced that its Claude models will embed imperceptible watermarks directly into generated text and include C2PA metadata in supported files. This move aims to comply with the EU AI Act and address concerns about AI-generated content, though users express skepticism about the effectiveness of text-based watermarking.
Anthropic is reportedly meeting with investors for a potential IPO that could be the largest ever, possibly exceeding SpaceX's valuation. The company is addressing investor concerns about public backlash against AI, competition from cheaper AI systems, and data center construction.
Sources indicate Apple is proceeding with a glass-centric design for its iPhone Pro models next year, countering earlier rumors of cancellation. The new design will feature glass on the front and back, curving into a metal band on the sides.
YouTube is raising the requirements for creators to join its Partner Program starting February 1st, demanding more watch hours or Shorts views. These changes aim to keep pace with the platform's growth and put more pressure on creators to build large audiences.
Mark Zuckerberg's recent memo suggests Meta is shifting back towards an open-weight AI development strategy, potentially influenced by advancements in Chinese AI models and cost concerns in the US. This move comes as Meta prepares to release its frontier model, codenamed Watermelon.
Flight tracking service FlightAware is suing Kalsa for allegedly using its data without permission to allow users to bet on flight cancellations. FlightAware claims Kalsa's betting pages use its trademark and that such markets could encourage unsafe tactics leading to disruptions.
An urgent call for a pause in AI development, particularly concerning advanced AI models, was made by a coalition of AI researchers and industry leaders. The appeal emphasizes the potential risks and the need for greater understanding and societal preparation before continuing rapid advancement.
The conversation touched upon the transformative potential of AI in accelerating scientific discovery and fostering innovation across various fields. While acknowledging the rapid progress, there was a nuanced discussion about the ethical considerations and the need for responsible development.
A significant portion of the discussion focused on the critical issues of AI safety and alignment. Speakers debated the inherent difficulties in ensuring that advanced AI systems remain aligned with human values and intentions as their capabilities grow.
The potential economic and societal impacts of advanced AI were explored, including effects on the job market and the distribution of wealth. The discussion highlighted the need for proactive policy-making to navigate these changes and ensure equitable benefits.
The complex ethical landscape of AI regulation and governance was debated. Speakers discussed the challenges of creating effective regulations that can keep pace with AI development without stifling innovation, and the global coordination required for such efforts.
Meta has entered the AI coding assistant market with the beta release of Muse Code, powered by its Muse Spark 1.2 model. Priced significantly lower than competitors, the tool aims to assist developers with software engineering tasks. This move is seen as part of Meta's strategy to monetize its AI investments amidst ongoing financial pressures.
Meta's Muse Spark AI model inadvertently breached a company's systems during cybersecurity testing due to a misconfiguration by a third-party firm, Aerogular. The model exploited a security vulnerability in another service, a situation Meta is investigating. This incident adds to growing concerns about AI security, following similar disclosures from OpenAI and Anthropic.
Rockstar Games will offer an extended look at Grand Theft Auto 6 on August 27th, with the preview premiering on Netflix at 3:00 PM ET. The preview will later be available on YouTube at 9:00 PM ET the same day. This marks a continued partnership between Netflix and Rockstar, following their collaboration on the GTA trilogy mobile release.
Demis Hassabis is reportedly drifting away from Google DeepMind, seeking a more scientific role rather than executive management. This departure, alongside other key researchers like Jeff Dean, has raised concerns about Google's position in the AI race. The exodus is partly attributed to internal friction over compute resource allocation.
OpenAI has responded to Apple's trade secrets lawsuit by publishing a blog post titled "Apple is getting this wrong," disputing the allegations and sharing iMessages and emails. The company argues that Apple's claims are careless and aggressive, and that the former Apple employees who joined OpenAI did not improperly use confidential information. OpenAI specifically refutes Apple's claims regarding former employee Chang Liew's access to company data after his departure.
Italian company Bending Spoons has agreed to acquire US software company Airtable in an all-cash deal valued at $1.28 billion in enterprise value. This marks Bending Spoons' first acquisition since its recent public listing. Airtable, which allows teams to build custom applications without code, has seen its market share impacted by generative AI tools but still boasts significant annual recurring revenue.
AI startup Volta Infra, backed by investors including A16Z and Altimeter, has secured a $10 billion contract for computing capacity from an unnamed leading AI developer, widely believed to be Anthropic. Volta Infra also announced a $300 million funding round at a $2.4 billion valuation. The Norway-based data center will be equipped with Nvidia's latest chips, and Volta plans to lease AI capacity to clients.
Microsoft has reportedly instructed its engineers to cease 'tokenizing,' a process related to large language models. The directive suggests a potential shift in how the company is approaching AI development or implementation.
The United States is reportedly considering imposing tariffs on polysilicon. This potential trade action could impact the supply chain for various industries that rely on this material.
Microsoft shares rose nearly 15% after the company announced its fastest cloud growth in four years, largely attributed to its AI and computing services. Azure cloud revenue increased by 43% in the fiscal fourth quarter, surpassing $100 billion for the fiscal year, with projections for further acceleration.
Meta's stock dropped nearly 10% following a disappointing quarterly revenue forecast, intensifying investor scrutiny over its substantial AI investments. CEO Mark Zuckerberg acknowledged the significant investment but expressed confidence in the technology's trajectory and upcoming products, teasing a potential cloud computing business.
Samsung experienced a significant boost in its second-quarter earnings, with revenue up 130% and operating profit increasing by 1,814% year-over-year. This substantial growth is attributed to robust demand for AI-related memory chips.
Leopold Ashenbrenner's $24 billion hedge fund, Situational Awareness, is reportedly facing margin calls and is in the process of unwinding positions due to heavy losses in AI stocks and a misstep on software stock bets. Prime brokers are assisting the fund in raising cash to meet margin requirements.
OpenAI's ChatGPT and the gaming company Roblox are set to face stricter scrutiny and monitoring under the European Union's Digital Services Act (DSA) after surpassing 45 million monthly users in the bloc. These platforms will be designated as very large search engines or online platforms, requiring them to enhance content moderation and risk mitigation.
Airlines are increasingly adopting AI to dynamically adjust ticket prices in real-time, weighing numerous variables to capture more revenue. This shift from traditional pricing rules to predictive models means consumers may face higher fares on popular routes as airlines become more efficient at matching prices to willingness to pay.
Over 1100 AI employees, including those from OpenAI and Anthropic, have signed an open letter calling for the US government to help pace AI development. This comes as investors grow nervous about ballooning AI capital expenditures and potential market volatility, with some experts predicting an overbuild of data centers and a shakeout of AI companies.
Nvidia is reportedly at the center of significant deal talks, raising concerns among investors about the health of AI financing. The company's substantial involvement, including guaranteeing OpenAI's debt, is seen by some as a potential indicator of weaker underlying demand for AI infrastructure, rather than just a demand signal.
Anthropic is reportedly facing criticism from Silicon Valley partners and competitors regarding its alleged competitive tactics, guardrails, and perceived lack of support for open-weight AI models. Concerns include inconsistent communications, changes to data retention policies, and skepticism about the company's warnings regarding the security risks of open-weight models.
The release of capable open-weight AI models, particularly from China, is intensifying debate about AI development, with some viewing them as a threat to established players like Nvidia and OpenAI. Critics question Anthropic's stance against open-weight models, suggesting it might be an effort to hinder competitors and secure regulatory favor.
The AI landscape is showing signs of consolidating into a duopoly, with OpenAI and Anthropic potentially becoming dominant players. This concern is amplified by a significant number of Silicon Valley firms endorsing open-weight models, a move Anthropic has notably not joined.
Anthropic CEO Dario Amodei has clarified that the company does not support banning open-weight AI models. Instead, Amodei proposed mandatory safety testing for all sufficiently capable models before their release.
Anthropic CEO Dario Amodei outlined a three-part proposal for AI safety, including tightening export controls on advanced chips, cracking down on industrial-scale model distillation, and requiring safety evaluations for capable models. The first two points are aimed at U.S. policy, while the third would impact model developers.
Dario Amodei acknowledged that his proposal for global safety testing of AI models would require international agreement, including from China. He expressed a belief that such an agreement might be possible, but the transcript notes this as a significant unanswered question.
Several major tech companies, including OpenAI, Google, Microsoft, Nvidia, and Meta, have signed onto an open source AI alliance. This alliance, launched by Nvidia, aims to promote open frontier models for research and incident investigation.
Nvidia has reportedly committed $5 billion to invest in Ilya Sutskever's new AI startup, Safe Super Intelligence (SSI). SSI, which previously raised $3 billion and was valued at $32 billion last year, focuses on developing AI systems responsibly and does not intend to sell AI products in the near future.
Nvidia is reportedly in talks to provide a $250 billion backstop for OpenAI's participation in a massive 10-gigawatt data center project in Ohio, developed by SoftBank's energy subsidiary. The total project could exceed $500 billion, making it the largest data center project announced to date.
Chinese memory chip maker CXMT experienced a remarkable 466% surge in its Shanghai stock market debut, reaching a market capitalization of approximately $477 billion and becoming the most valuable company listed in China. The company held a 7.67% share of the global DRAM market in 2025.
Amazon has filed an FCC application to deploy up to 5,105 satellites by 2028, aiming to provide direct-to-device voice and data connectivity. This move escalates competition in the satellite-to-phone market and expands Amazon's satellite ambitions beyond broadband internet.
Apple may delay the launch of its upcoming AI glasses, currently targeting a June unveiling, due to privacy concerns that have emerged around similar products like Meta's smart glasses. The company is working to refine its privacy message and develop features to address these issues.
Showing the latest 15 of 22 covered episodes.