★ Top Story
The VergeOpenAI CEO Sam Altman has floated giving the US government a 5% ownership stake in the company as a way to ease tensions with the Trump administration and soften public backlash against AI. The proposal, reportedly discussed with the White House, is framed as a way to give ordinary Americans a financial stake in the AI boom. Critics note the offer falls well short of the public ownership share some lawmakers have demanded.
Read the original →
TechCrunchAt an internal all-hands meeting, Meta CEO Mark Zuckerberg told employees that the company's AI agent development has not advanced as quickly as he had anticipated. The candid admission is notable given Meta's heavy investment in AI infrastructure and its public ambitions to lead in agentic AI. It also echoes broader industry signals that autonomous AI agents remain harder to deploy reliably than early optimism suggested.
Read more →
TechCrunchMicrosoft has stood up a dedicated AI deployment subsidiary backed by a $2.5 billion commitment, joining Amazon, OpenAI, and Anthropic in building out enterprise AI delivery arms. The new unit is designed to accelerate how businesses adopt and operationalize AI at scale. The move signals that the major tech players are shifting focus from model development toward capturing revenue from real-world AI deployments.
Read more →
TechCrunchAnthropic is reportedly in discussions with Samsung to co-develop a custom AI chip, following OpenAI's recently announced chip partnership with Broadcom. The move reflects a growing trend among frontier AI labs to reduce dependence on Nvidia by building proprietary silicon tailored to their specific inference and training workloads. A successful deal would give Anthropic greater control over its compute costs and supply chain.
Read more →
Ars TechnicaGoogle's aggressive expansion of AI data centers caused its total electricity consumption to jump 37% in 2025, according to the company's latest environmental report. The dramatic increase underscores the enormous energy cost of scaling large language models and inference infrastructure. Google says it is working to offset the surge through clean energy procurement, but the gap between AI demand and sustainable supply remains a serious challenge.
Read more →
TechCrunchMeta has soft-launched Pocket, an experimental app that lets users generate and share playable mini-games simply by typing text prompts — a so-called "vibe coding" approach to game creation. The app represents Meta's latest attempt to find consumer AI products beyond its core social platforms. It also signals the company's interest in AI-generated interactive content as a new entertainment category.
Read more →
TechCrunchCloudflare has issued a new policy requiring AI companies to clearly distinguish between web crawlers used for search indexing and those used for AI training or agent tasks by September 15, or face being blocked by default across a large swath of publisher sites. The move is a significant escalation in the ongoing battle over whether AI companies should compensate content creators for training data. It gives Cloudflare's massive network of publisher clients a practical enforcement mechanism against unauthorized AI scraping.
Read more →
MIT Technology ReviewA new MIT Technology Review investigation highlights a persistent and underappreciated flaw in today's leading large language models: they tend to converge on the same outputs, producing homogenized responses that lack genuine diversity of thought. A startup is developing techniques to break this "groupthink groove" by introducing structured variation into how models reason and respond. The problem has real consequences for use cases that depend on creative, exploratory, or adversarial thinking.
Read more →
TechCrunchVenice AI has raised a $65 million Series A at a valuation that pushes it into unicorn territory, fueled by strong demand for its privacy-preserving AI platform. The company is already profitable, with annualized revenues exceeding $70 million, according to CEO Erik Voorhees. Venice's growth reflects a meaningful market segment of users and enterprises who want AI capabilities without surrendering their data to centralized model providers.
Read more →
MIT Technology ReviewAnthropic unveiled Claude Science at an event targeting pharmaceutical executives, biotech founders, and academic researchers, positioning it as the company's most capable product for scientific discovery and analysis. The launch marks a strategic push by Anthropic to capture high-value professional markets where AI-assisted research could command premium pricing. It also intensifies competition with Google DeepMind and OpenAI, both of which have made scientific AI a priority.
Read more →
TechCrunchMeta is developing plans to monetize its surplus AI compute capacity by offering cloud infrastructure services, putting it in direct competition with AWS, Google Cloud, and Microsoft Azure. The strategy mirrors a similar move by SpaceX, which has also been exploring ways to commercialize its AI hardware investments. If executed, Meta's entry into the cloud compute market would represent a major structural shift in the competitive landscape for AI infrastructure.
Read more →
TechCrunchGoogle has brought Gemini Spark, its always-on agentic AI assistant, to macOS, expanding beyond its initial platform availability. The update also includes real-time tracking capabilities and support for a broader range of third-party apps. The Mac launch puts Gemini Spark in more direct competition with Apple Intelligence and other desktop AI assistants vying for the role of primary productivity layer on personal computers.
Read more →
TechCrunchSpaceX reportedly showed investors a prototype "handset-like" AI device ahead of a potential public offering, fueling speculation that the company is planning to enter the consumer hardware and wireless market. The device's AI-first design would position it alongside other emerging AI hardware plays like the Humane Pin and Rabbit R1, though SpaceX's Starlink connectivity could give it a meaningful differentiator. The move suggests Elon Musk's ambitions for SpaceX extend well beyond rockets and satellite internet.
Read more →
Ars TechnicaPrivacy advocacy groups have filed warnings with the Federal Trade Commission urging regulators to reject Elon Musk's bid to end court-ordered monitoring of X, citing the platform's AI data practices as a particular concern. The advocates argue that X's use of user data to train Grok and other AI systems creates significant privacy risks that warrant continued oversight. The filing adds regulatory pressure to X at a moment when Musk's influence over federal agencies is already under intense scrutiny.
Read more →
TechCrunchA TechCrunch analysis of Jersey Mike's IPO documents found the sandwich chain invoking AI despite having no meaningful AI business — a telling sign of how thoroughly the term has been co-opted as a buzzword in public market filings. The phenomenon mirrors the blockchain hype cycle of the late 2010s, when companies across unrelated industries added crypto language to boost valuations. It raises questions about whether investors and regulators are scrutinizing AI claims with sufficient rigor.
Read more →
NVIDIA BlogNVIDIA has announced a new initiative inviting capital partners to co-invest in large-scale, multi-tenant AI compute infrastructure designed to support token-scale inference services. The program is aimed at emerging AI companies that need access to significant GPU capacity quickly but lack the capital to build their own data centers. The move positions NVIDIA not just as a chip supplier but as a central orchestrator of the broader AI infrastructure financing ecosystem.
Read more →No stories from this source today.