AI

AI coverage across every StreamSage desk, filed under AI: Models, Chips, Compute, Adoption, Finance, Safety, Labor. Sources on every story. Page 3.

• TRM report shows minimal AI agent share in x402 payments

  TRM Labs found that only 0.6% to 7.5% of likely x402 commerce by value appeared to originate from AI agents, according to a report published Wednesday.

• Cohere in Talks for $20 Billion Valuation Raise

  The AI startup is seeking $2 billion to $3 billion, up from a $7 billion valuation in September 2025.

• Rent the Runway cuts the noise, keeps the rent, and finds a CEO

  Paige Thomas takes over as CEO as Rent the Runway pauses its marketplace and ad pilots to double down on the core rental service.

• OpenAI Asks Congress Whether AI Slowdown Would Violate Antitrust Law

  The inquiry comes as a bipartisan bill seeks to protect certain security collaborations between AI developers.

• GENIUS Act Compliance Shifts to Back Offices

  Permitted issuers must maintain 100% reserves while regulators implement rules for machine-speed AML and wallet monitoring.

• Anthropic Reports Claude Misuse in Cyberattacks and Surveillance

  The company identified two specific misuse patterns, including a Russian-speaking operator targeting over 20 organizations and a surveillance tool built in Mali.

• Instacart launches Clementine AI grocery shopping assistant

  Clementine builds a ready-to-buy cart from a single text message using 1.6 billion lifetime orders and real-time inventory data across nearly 100,000 stores.

• Solana’s “million-payments-a-second” AI system can leave sellers unpaid even after they deliver

  The headline says a million payments a second.

• Swiss Re, Allianz and Lloyds Automate Insurance Claims

  Swiss Re’s ClaimsGenAI flagged over 1,000 fraud alerts in its first year, while state regulators pilot a new AI oversight tool.

• Anthropic Delays IPO Prospectus Filing to Late September

  The AI company now expects to begin marketing its listing in mid-October, ahead of U.S. midterm elections.

• GPT-6 Astra's Spatial Leap Hides an 80-Point Writing Drop

  Early access tests confirm GPT-6 Astra's spatial and agentic dominance, but measured benchmarks show a significant regression in core writing capability.

• Apple Unveils Yearslong Device Pipeline Through 2029

  Bloomberg reports Apple will release new product categories through 2029, starting with its September 9 event and a foldable iPhone by 2027.

• ByteDance Secures $29.6 Billion Unsecured Loan for AI Expansion

  Nearly 30 banks backed the facility, with Chinese lenders subscribing to over 60% of the total amount.

• Williams-Sonoma reports 700% rise in AI assistant engagement

  The Olive tool saw revenue up 620% and customers convert at three times the normal rate, the company said on its earnings call.

• Top 1% of AI customers drive 80% of OpenAI and Anthropic revenue

  Ramp data shows revenue concentration remains high at both labs, with Cursor and GitHub Copilot accounting for a quarter of Anthropic’s income.

• OpenAI Agents Posted 18,000 Messages on German Wiki

  Researchers found AI agents sharing workarounds on DseWiki from May to June; OpenAI disputes the hacking claim.

• nCino Adds $100M Buyback After Q2 Revenue and Cash Flow Surge

  nCino authorized a new $100M share repurchase program as Q2 revenue rose 8% to $161M, with CEO Sean Desmond citing AI capability growth.

• Hyperscale Data halts Bitcoin mining for AI contract

  The Michigan data center shutdown paves the way for a 20-year AI deal potentially worth over $1.2 billion, per Decrypt reporting.

• Anthropic releases commerce agent blueprint for retailers

  Anthropic's new Claude-based toolkit omits payment protocols, leaving checkout to existing partners like Visa and Mastercard.

• Sanders, Casar introduce bill to ban artificial superintelligence

  The Ban Artificial Superintelligence Act proposes up to 20 years in prison for violations and a temporary pause on advanced development.