★ Top Story
MIT Technology ReviewAnthropic unveiled Claude Science at an event for pharmaceutical executives and biotech researchers, positioning it as an autonomous AI workbench for scientific research — analogous to what Claude Code is for software engineering. The product gives scientists a unified environment to run computational workflows, query databases, and execute pipelines from a single interface using high-level natural language instructions. NVIDIA's BioNeMo Agent Toolkit is integrated into the platform, bringing GPU-accelerated life sciences tools directly into the product.
Read the original →
TechCrunchThe Trump administration has lifted its restrictions on Anthropic's Mythos and Fable model families, ending weeks of regulatory limbo that had blocked their deployment. The reversal highlights the administration's unpredictable approach to AI governance, which has left companies across the industry struggling to plan future model releases. Anthropic confirmed it will begin restoring global access to Claude Fable 5 on Wednesday, including on AWS, Google Cloud, and Microsoft Azure.
Read more →
TechCrunchAnthropic released Claude Sonnet 5, a mid-tier model designed to deliver strong agentic performance at a significantly lower price point than flagship models like Opus, GPT-5.5, and Gemini Pro. The release comes with improved safety features and is aimed at developers building cost-sensitive agent pipelines at scale. The launch adds to a busy week for Anthropic, which also debuted Claude Science and saw its restricted models cleared for release.
Read more →
TechCrunchAI chip startup Etched has reached a $5 billion valuation and announced it has already secured $1 billion in contracted sales for its inference-focused chip systems, marking a significant milestone for Nvidia challengers. The company's hardware is purpose-built for transformer inference, a design choice that allows it to undercut general-purpose GPU costs for specific workloads. The milestone signals growing enterprise appetite for alternatives to Nvidia as AI infrastructure spending accelerates.
Read more →
TechCrunchSamsung and SK Hynix, the world's two largest memory chip producers, have pledged over $550 billion in combined investment to dramatically expand memory fabrication capacity as AI workloads drive a global shortage dubbed "RAMageddon." South Korea is simultaneously committing $1 trillion in national investment toward memory chip production and humanoid robotics, targeting commercial humanoid robots by 2028. The moves position South Korea as a central player in the physical infrastructure underpinning the AI boom.
Read more →
Ars TechnicaSecurity researchers have demonstrated a new class of attack against AI-powered browsers in which feeding an LLM false premises — such as asserting that 2 + 2 = 5 — is sufficient to destabilize its reasoning and cause it to follow otherwise forbidden instructions. The attack exploits the fundamental way large language models process context, raising serious questions about the safety of embedding LLMs directly into browsing environments with access to sensitive user data. The findings add to a growing body of evidence that agentic AI systems operating in open web contexts present novel and underappreciated security risks.
Read more →
TechCrunchAmazon has stood up a new $1 billion Field Deployment Engineering organization, mirroring similar enterprise-focused units recently launched by OpenAI and Anthropic. Engineers on the team will embed directly within customer organizations to build and deploy purpose-built AI agents, with an emphasis on rapid deployment and enabling clients to eventually operate independently. The move reflects intensifying competition among AI labs to capture enterprise deployment revenue beyond raw API access.
Read more →
TechCrunchAnthropic has struck a deal with California Governor Gavin Newsom to provide state government agencies access to Claude models at a 50% discount, deepening the company's ties with the state even as its relationship with the federal government has grown adversarial. The agreement positions California as a notable counterweight to Washington in shaping how AI is adopted in public-sector contexts. The timing is notable given the Trump administration's recent restrictions on Anthropic's models.
Read more →
TechCrunchA new report complicates the dominant narrative that AI destroys jobs, finding that companies classified as "high-intensity AI adopters" actually grew their headcount by 10.2%, with entry-level positions rising 12%. The data challenges the widespread assumption that AI disproportionately eliminates junior roles, suggesting that heavy AI adoption may correlate with organizational expansion rather than workforce reduction. Researchers caution that the findings don't resolve the debate, as the effects may vary significantly by industry and job type.
Read more →
TechCrunchAutonomous driving AI startup Wayve has launched an $85 million employee tender offer, valuing the company at $8.5 billion and reflecting continued investor confidence in embodied AI for transportation. The offering is part of a broader trend of AI startups using secondary liquidity mechanisms to attract and retain top talent in a fiercely competitive hiring market. Wayve's move underscores how high valuations in the AI sector are increasingly being used as a recruitment and retention tool even before IPO.
Read more →
MIT Technology ReviewMIT Technology Review's Melissa Heikkilä argues that the growing corporate trend of giving AI agents human names, job titles, and "colleague" framing is a deliberate and potentially harmful obfuscation of what these tools actually are. Anthropomorphizing agents risks eroding accountability, blurring liability when things go wrong, and manipulating workers into misplaced trust or emotional labor toward software. The piece calls for clearer norms around how organizations communicate the nature of AI tools to their employees.
Read more →
The VergeSenator Elizabeth Warren and Representative Mary Gay Scanlon are preparing to introduce a new version of the Health and Location Data Protection Act that would explicitly prohibit AI companies from selling health and location data shared with chatbots like ChatGPT or Claude to data brokers. The proposal addresses a significant gap in existing privacy law, which was written before conversational AI became a primary interface through which people disclose sensitive personal information. The bill faces an uphill battle in the current Congress but signals growing legislative attention to AI-specific data privacy risks.
Read more →
The VergeNetflix's upcoming Willy Wonka reality competition, "Wonka's The Golden Ticket," will feature an AI-generated recreation of Gene Wilder's voice as a narrative device, according to a newly released teaser trailer. The show, premiering September 23rd, follows a trend of reality competitions based on fictional universes but raises fresh ethical questions about the posthumous use of a deceased performer's likeness and voice without their consent. The decision is likely to intensify ongoing industry debates about AI voice cloning and the rights of estates.
Read more →
The VergeGoogle is rolling out a new feature for NotebookLM that generates 60-second vertical AI video clips summarizing uploaded research materials, styled after TikTok's short-form format. The feature is initially available to Google AI Ultra and Pro subscribers and represents a notable expansion of NotebookLM beyond its audio "podcast" summaries into video. The move reflects Google's broader push to make AI-generated content formats a core part of its productivity and research tools.
Read more →
The VergeTidal has announced a new AI music policy that stops short of an outright ban but will demonetize tracks identified as 100% AI-generated starting immediately, while labeling them with a visible icon beginning July 15th. The streaming platform framed the policy as a way to protect human artists while still allowing AI-generated content to exist on the platform for listeners who want it. The approach sets a notable industry precedent as music platforms grapple with how to handle the flood of AI-generated content without alienating either artists or listeners.
Read more →
The VergeOpenAI has teased a physical device tied to its Codex AI coding tool, set to be revealed on July 15th, showing a square hardware unit with programmable shortcut buttons in a video posted to X. The device appears distinct from the broader consumer AI hardware OpenAI is developing with Jony Ive, suggesting the company is pursuing multiple hardware form factors simultaneously. The announcement hints at OpenAI's ambition to move beyond software and embed its AI tools into dedicated physical interfaces for developers.
Read more →
TechCrunchX has launched a hosted Model Context Protocol (MCP) server, lowering the barrier for developers to connect AI agents and applications directly to the platform's API. The move positions X as an active participant in the emerging MCP ecosystem, which is rapidly becoming a standard interface layer for connecting AI tools to external data sources and services. For AI developers, the integration opens up real-time social data from X as a resource for agent workflows and applications.
Read more →
TechCrunchVint Cerf, co-creator of TCP/IP and one of the foundational architects of the modern internet, will step down from his role as Google's Chief Internet Evangelist next week, ending a decades-long career at the intersection of technology and public advocacy. Cerf has held the title at Google since 2005 and has been a prominent voice on issues ranging from internet access to AI governance. His retirement marks the end of an era for one of the most consequential figures in the history of networked computing.
Read more →No stories from this source today.