the Microdose

OpenAI is 76% Terrifying

+ China dumps Nvidia, AI owes you, and deepfake insurance
Adam Wildheart
retro 80s computer with OpenAI logo surrounded by pixelated warning signs
retro 80s computer with OpenAI logo surrounded by pixelated warning signs

The Microdose AI

Subscribe to The Microdose
Subscribe -- post
Cheri Wildheart
Adam Wildheart

Good morning. You know those holiday ads that accidentally become classics? McDonald’s just dropped an AI-generated commercial so spectacularly bad it might earn cult status. The 45-second slop spot joked that the holidays are “the most terrible time of the year.” People hated it, calling out the weird visuals and creepy characters. Backlash was so intense McDonald’s quickly shut down comments and yanked the ad offline. Their ad agency claims it took 7 weeks to make this “cinematic film,” but you’d never guess by watching it. 

OpenAI’s latest AI models are so smart, it’s practically criminal. They just confessed that upcoming models pose a “high” cybersecurity risk, capable of launching zero-day exploits and automating sophisticated attacks. GPT-5 already scored 27% on a capture-the-flag cybersecurity test. Its newer cousin, GPT-5.1 Codex Max, nailed a terrifying 76%. OpenAI warns future models will be even scarier, prompting them to rush out defensive tools and launch a special “Frontier Risk Council” to prevent their creations from going rogue. Usually it’s Anthropic yelling about doom. When OpenAI joins in, it can’t be good.

China shuns Nvidia by mandating domestic AI chips. Beijing has given a clear message to the US that their AI chips aren’t welcome. State institutions and universities have strict orders to rely exclusively on homegrown tech. Domestic processors still can’t match the powerful H200 in speed or efficiency, so China is subsidizing electricity for data centers running slower local hardware. Privately authorities are meeting with Alibaba and Tencent to gauge their needs, hinting they might still grant special approval. Sorry Nvidia, China still prefers black market chips

🤜🤛 TOGETHER WITH FLOW

Typing is tiring. With Wispr Flow, you just speak.

Flow connects your natural voice to tools like ChatGPT or Claude, turning your long, detailed prompts into clean text. On the go or at your desk, it enables you get more done, faster.

Save hours a day. Get Flowing

ChatGPT’s got Adobe Photoshop and Acrobat. Now you can blur backgrounds and edit images directly inside a chat, no license required. Need a PDF edited? ChatGPT merges files, extracts data, and rewrites text effortlessly. Adobe saw the writing on the wall. With Canva and Figma already embedded in ChatGPT, they had to jump fast or risk irrelevance. Welcome to the future, where even premium pro tools kiss OpenAI’s ring.

The world would be a better place if publishers got paid by AI scrapers. Now you can set clear rules for how AI companies scrape and pay for your content, thanks to a new licensing standard called RSL 1.0 (Really Simple Licensing). If you use Cloudflare, sign up for their Pay Per Crawl beta and they’ll handle all enforcement and payments. Simply set your price and watch the money roll in. AI scrapers keep scraping, publishers (finally) get paid, and Google has a blueprint for playing fair. Almost sounds civilized.

AI deepfakes are stealing your face, and there’s not much you can do about it. Hyper-realistic AI deepfakes aren’t just trolling celebrities. They’re disrupting elections, stealing millions of dollars, and exploiting people in deeply disturbing ways. Denmark wants to make it illegal to use your face or voice without your permission. In the US, you can’t trademark your likeness. But businesses can insure it. Coalition just launched a Deepfake Response Endorsement that treats your identity like a SaaS product, sending in lawyers, forensic teams, and PR experts when things go sideways. Nothing more American than outsourcing your dignity to tech support.

fun stats

🚀 $1.5 trillion. Valuation target for SpaceX’s rumored 2026 IPO. If it lands, this would make it the largest IPO of all time.

🎁 $35 billion. Amount Amazon plans to invest in India by 2030. Tech giants Microsoft ($17.5B) and Google ($15B) are also investing heavily to boost India’s AI ecosystem. 

💡 57%. Share of all AI agent activity focused on cognitive work such as productivity and learning rather than simple automation tasks. Turns out agents are more “thinking partners” than AI butlers.

subscribe today

Get an edge with The Microdose

Skip the prompts, get the signal. We’ll send you short daily updates about the real AI + future tech you need to know. Fast, smarter, mildly addictive – and free. 

Subscribe -- page