Anthropic's annualized revenue surged to $47 billion in May, up from roughly $9 billion at the end of 2025, as the company prepares for a public offering. Daniela Amodei pushed back on skeptics questioning whether AI investment will generate real returns, arguing the growth trajectory speaks for itself. The IPO will be a major test of whether public markets share that confidence.
Read the original →TSMC CEO C.C. Wei acknowledged that surging demand from American AI customers is outpacing the chipmaker's capacity, even as it expands its US manufacturing footprint. The bottleneck underscores how the global semiconductor supply chain remains a critical constraint on AI infrastructure buildout. The admission signals that chip scarcity could persist as a limiting factor for AI development well into the near future.
Read more →In a rare show of unity, competing AI companies signed an open letter urging Congress to close what they describe as a dangerous biosecurity gap that could allow bad actors to use AI models to develop biological weapons. The letter calls for specific legislative guardrails rather than voluntary industry commitments. The move is notable given the fierce rivalry among signatories, suggesting the threat is viewed as serious enough to warrant collective action.
Read more →Federal judges are increasingly confronting a wave of AI-drafted legal filings from pro se litigants, raising serious questions about accuracy, hallucinated citations, and due process. Judge Maritza Braswell in Colorado described sifting daily through AI-generated documents that range from coherent to deeply flawed. The trend is forcing courts to develop new protocols for identifying and handling AI-assisted filings before they clog dockets or mislead judges.
Read more →Critics are warning that the Trump administration's new executive order on AI safety testing is largely performative, given that DOGE-driven cuts have hollowed out the federal agencies responsible for carrying out such evaluations. Without adequate staffing and technical expertise in government, mandated model testing may amount to little more than a paper exercise. The gap between policy ambition and institutional capacity could leave dangerous AI deployments effectively unchecked.
Read more →UK regulators have directed Google to improve source attribution in its AI Overviews and provide a mechanism for publishers to opt out of having their content used in generative search results. The order comes after Google argued that users don't want to see extensive source lists — a position regulators rejected. The opt-out tool is set to be piloted in the UK before a broader global rollout, marking a significant precedent for AI search regulation.
Read more →Alphabet completed an $85 billion stock sale tied to its AI business, setting a new record for capital raises in the sector and demonstrating that institutional investors remain deeply bullish on AI's commercial future. The raise dwarfs previous tech capital events and provides Google with a massive war chest to compete in the AI infrastructure and model race. Analysts say the sheer scale of investor demand suggests the market has not yet priced in a ceiling for AI growth.
Read more →Amazon unveiled a new generation of its Proteus warehouse robot that accepts natural language instructions from human workers, replacing the need for specialized programming or code-based commands. The upgrade is part of Amazon's accelerating push to automate its fulfillment operations as it scales back its human workforce. The language interface is designed to lower the barrier for workers to direct robots on the fly, though labor advocates note the broader automation trend raises displacement concerns.
Read more →The Estonian government has published a benchmark evaluating dozens of large language models on their ability to identify and resist Russian "strategic narratives" — state-sponsored disinformation framing. The results show significant variation across models, with some proving far more susceptible to amplifying propaganda than others. The benchmark is the first government-led evaluation of its kind and could inform procurement decisions for public-sector AI deployments across NATO countries.
Read more →Meta is deploying temporary tent structures to house data center equipment, borrowing a construction shortcut previously used by Tesla to accelerate factory buildouts. The approach allows Meta to bring AI compute capacity online faster and at lower upfront cost than traditional purpose-built facilities. As AI infrastructure spending reaches eye-watering levels across the industry, the tactic reflects growing pressure on hyperscalers to find creative ways to manage capital expenditure.
Read more →Former OpenAI CTO Mira Murati is re-emerging publicly after a period of deliberate quiet following her high-profile departure, signaling that her new AI venture is approaching a stage where visibility matters. The piece examines how founders in stealth mode must eventually make noise to attract talent, investors, and customers — even when they'd prefer to stay heads down. Murati's careful re-entry is being watched closely given her stature in the AI field.
Read more →Airbnb CEO Brian Chesky announced plans to establish a dedicated AI research lab, a significant strategic pivot for a company that has until now held off on major LLM partnerships, citing the immaturity of existing products. The lab signals Chesky's intent to build proprietary AI capabilities rather than rely entirely on third-party models. The move puts Airbnb among a growing list of non-AI-native companies investing directly in foundational AI research.
Read more →A new analysis argues that viral robot demonstration videos are systematically distorting public understanding of where humanoid robotics actually stands, with carefully staged clips masking significant limitations in real-world reliability and generalization. The piece breaks down common techniques used in demos — cherry-picked takes, controlled environments, and selective editing — that make robots appear far more capable than they are in practice. As investment in humanoid robotics surges, the gap between perception and reality has meaningful consequences for both funding and regulation.
Read more →Startup Poke has secured Apple's approval to operate as the first AI agent on the Messages for Business platform, enabling businesses to deploy conversational AI directly within Apple's native messaging environment. The milestone opens a significant new distribution channel for AI agents, bypassing the need for users to download separate apps. It also marks Apple's first meaningful step toward allowing third-party AI agents into its tightly controlled messaging ecosystem.
Read more →Apple's Worldwide Developers Conference is approaching with expectations running high for a substantial overhaul of Siri, which has lagged behind competitors like Google Gemini and OpenAI's voice interfaces. Leaks and analyst reports suggest Apple Intelligence will receive significant new agentic capabilities, tighter on-device processing, and deeper third-party app integration. The announcements will be a critical test of whether Apple can close the gap with rivals who have moved faster on conversational AI.
Read more →NVIDIA CEO Jensen Huang traveled to Seoul this week to deepen partnerships with South Korean companies across sovereign AI infrastructure, robotics, and gaming. South Korea has emerged as one of the world's most active AI ecosystems, with major investments in national AI compute capacity and a dense network of hardware and software partners. The visit underscores NVIDIA's strategy of anchoring sovereign AI buildouts around its platform globally.
Read more →The Verge's editors reflect on hands-on time with Google's new Gemini-powered agent Spark, which demonstrated an unsettling ability to surface deeply personal details — a user's dog's name, a spouse's first name — without being explicitly told them, by mining connected account data. The piece argues that as AI agents become genuinely effective, they expose a foundational tension: the more useful they are, the more they depend on surveillance-scale data access. The editorial frames this as a broken promise at the heart of the AI assistant paradigm.
Read more →Summaries are AI-generated. Click through to read the original reporting.
Summaries are AI-generated. Click through to read the original reporting.