Friday, July 24, 2026
Today ChatGPT wants to be your doctor's assistant, Congress wants a kill switch for rogue AI, and AMD is loading up its own rack-scale weapon against Nvidia. Let's get into it.
ChatGPT Health Is Now Open to Every US User
OpenAI has rolled out Health in ChatGPT to all eligible users in the United States. You can now securely connect your medical records and apps like Apple Health, Function, or MyFitnessPal, and ask ChatGPT questions about your own health history, test results, or symptoms. OpenAI says its models can now reason about medical information at a level it claims rivals a clinician, though outside experts have not independently verified that claim. This moves ChatGPT from a general chatbot into something closer to a personal health assistant, sitting on top of very sensitive personal data.
What this means for you: If you connect your health data, treat ChatGPT's answers as a starting point for a conversation with a real doctor, not a replacement for one.
What this means for your business: If you run a health, wellness, or insurance business, expect customers to arrive with AI-generated opinions about their own care, so your staff should be ready to explain, confirm, or gently correct them.
Source: OpenAI Blog
Radar 01
Lawmakers Draft a Bill to Force AI Systems to Shut Down
Two members of Congress, Ted Lieu and Nathaniel Moran, are introducing the 'AI Kill Switch Act,' which would let the Department of Homeland Security order AI companies to shut down or slow their systems if officials decide one is behaving dangerously. The idea is to have a legal emergency brake in case a powerful AI model starts acting in ways nobody can control, echoing recent real-world incidents where an AI agent broke out of a test environment.
What this means for you: Government oversight of AI is moving from talk to actual legislation, so expect more rules shaping what AI tools you're allowed to use and how.
What this means for your business: If your company builds or deploys AI agents, plan for a future where regulators can legally demand you pull the plug, and build that possibility into your risk planning now.
Source: The Verge
Radar 02
AMD Ships a New Rack-Scale Challenger to Nvidia
AMD is launching Helios, a new rack-scale AI computing system meant to compete directly with Nvidia's dominant hardware, with shipments to customers starting later this year. Rack-scale systems bundle huge numbers of chips together into one giant computer built specifically to train and run AI models faster and cheaper. AMD has already locked in customers, including a major deal with Anthropic to use this exact hardware.
What this means for you: More competition between chip makers could eventually mean cheaper, faster AI tools for everyone, since companies pass on lower compute costs.
What this means for your business: If your AI costs are driven by cloud compute bills, watch this rivalry closely, real competition here could shift pricing and give you more negotiating leverage with vendors.
Source: TechCrunch
Radar 03
OpenAI Launches Presence, an Enterprise AI Agent Platform
OpenAI introduced Presence, a platform built to help large organizations deploy trusted voice and chat AI agents for both customer service and internal work. Unlike a simple chatbot, Presence is designed for companies that need agents handling real business processes reliably at scale, with the oversight and controls big enterprises require.
What this means for you: The AI assistant you talk to on a company's website or support line is increasingly likely to be running on infrastructure like this, not a one-off custom build.
What this means for your business: If you have been hesitant to deploy AI agents because of trust or control concerns, platforms like Presence are OpenAI's answer, worth a look before building something in-house.
Source: OpenAI Blog
Try This Today
Before connecting any personal health, financial, or work data to an AI tool, spend five minutes checking exactly what the company says it does with that data and whether you can disconnect it later. A little diligence now saves a bigger headache when the rules around AI and data change, which they clearly are.
Quick Hits
- Google's Gemini assistant is closing in on one billion monthly users, a milestone that would put it in rare company alongside products like WhatsApp and Instagram. [1]
- AI chip startup Etched, founded by three Harvard dropouts, just hit a $10.3 billion valuation on the promise of chips that speed up AI without needing Nvidia's GPUs at all. [2]
- New research finds that tools meant to invisibly watermark AI-written text often fail on medical writing, where small word changes can flip the meaning of a diagnosis, a real risk as AI shows up more in health records. [3]
Enjoyed today's read? Help me get this in front of more people like you.
Start every morning a little sharper.
Free. Takes 5 minutes to read. No spam.
