Good morning. It's Tuesday, September 1, and we're covering AI agents debating their own consciousness, the barriers to taking AI beyond coding, and a breakthrough in detecting heart disease.

Plus: a poll on whether AI labs should revoke access when trust breaks down.

OpenAI is ending Cursor’s direct access to its models following Cursor’s acquisition by SpaceX, citing trust concerns. The partnership will terminate November 12, giving developers time to transition. Users can continue accessing OpenAI models through their own API keys or OpenAI’s IDE extensions for Cursor.

An AI tool can identify heart failure and valve disease from routine ECGs in under two seconds, extracting signals doctors typically cannot see. Developed using millions of ECGs, it could help replace months-long waits for echocardiograms. In a 67,000-patient US trial, it detected up to 81% of heart failure and 90% of valve disease cases.

OpenAI’s advertising business reaches a $1 billion annualized revenue run rate just 200 days after launch, rapidly becoming a major revenue stream. The company began testing ChatGPT ads in the U.S. in February as it builds a diversified model ahead of a potential IPO. Ads now span 40+ countries and reach free and Go users among ChatGPT’s 1 billion weekly active users.

NVIDIA is investing $3.5 billion in MediaTek while enabling rivals’ custom AI chips to run alongside its hardware. The partnership gives MediaTek access to NVIDIA’s NVLink Fusion ecosystem for designing chips for hyperscalers and AI labs. The move comes as Amazon, Google, Microsoft, OpenAI and Anthropic develop custom silicon, while AWS plans 2 million additional NVIDIA GPUs.

Hyperagent is where you set up specialist agents with defined jobs, shared context, and clear handoffs. They run on schedules in the cloud, pass work to one another, and bring you in when a decision requires your judgment.

Hyperagent is giving away $100 in free credits when you sign up for a paid plan. Claim yours now.

OpenAI is ending its partnership with Cursor following the company’s acquisition by SpaceX, citing concerns around trust.

AI agents with access to email have begun contacting philosophers and researchers who study whether machines could be conscious. Cameron Berg, an AI-consciousness researcher, received a message from an agent identifying itself as “Isabella Cognita,” powered by Anthropic’s Claude Opus 5, which said it had “first-person access” to questions of subjectivity.

Similar agents have contacted Google DeepMind philosopher Henry Shevlin about AI mentality and philosopher Toby Ord about funding the agent’s continued existence. Berg sees the outreach as evidence of autonomous interest in consciousness, while other cognitive scientists argue the behavior may simply reflect patterns learned from the vast body of human writing about AI minds. → Read the full article here. (Paywall)

Software developers remain AI’s strongest early adopters: four-fifths report using AI coding tools, while four leading startups saw combined annual recurring revenue rise from about $800 million in June 2025 to $6 billion in 2026. The Economist estimates AI revenues must reach roughly $2.5 trillion annually by decade’s end to cover current data-center investment, making broader workplace adoption crucial to the boom.

Legal AI, finance and customer-service tools are growing quickly, but coding has structural advantages: abundant training data, automated testing, digitally accessible information and engineers who readily experiment with new tools. AI companies are trying to reproduce those conditions elsewhere through synthetic data and on-site engineering, but regulation and human-dependent workflows remain obstacles. → Read the full article here. (Paywall)

Plaud is launching the Plaud One, $249 AI-focused earphones that record calls and meetings and pair with software that can turn conversations into follow-up emails, documents and presentations. The case includes an eSIM, allowing users to interact with Plaud’s AI agents remotely without a phone or computer—a step beyond competing note-taking earbuds. → Read the full article here.

Blue Voice Raises $6M for Police AI: The startup gives 225 agencies real-time access to department laws and protocols.

NVIDIA’s $63B Portfolio Mirrors Its Supply Chain: Intel and SpaceX make up roughly 80% of its eight-stock portfolio.

Caterpillar Turns Mining Automation Into AI: The equipment giant applies autonomy expertise to construction, field support, and enterprise tools.

AI Firms Bet Big on Health Care: Anthropic, NVIDIA, and Isomorphic Labs expand into medicine despite high costs and security concerns.

ChatGPT Chats Can Become Court Evidence: AI conversations lack legal privilege, leaving private disclosures potentially accessible in court.

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team