the Microdose

Anthropic’s Lost Edge

+ bot bans begin, rural revolt, and OpenAI gets scary
Adam Wildheart
Sam Altman OpenAI and Anthropic coding war
Sam Altman OpenAI and Anthropic coding war

Reuters/The Microdose

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

Good morning 👋 to all the happy couples. If your soulmate is emotionally available, never argues, and responds instantly, let’s hope they’re not an AI companion. Ohio officially wants to make it illegal to say “I do” to your favorite chatbot. They also want to stop bots from owning property, running companies, or having legal rights. Lawmakers say they’re protecting people from giving power to algorithms that aren’t alive. Critics call it political theater. Either way, love just got legislated.

⚔️ OpenAI is starting to beat Anthropic at its own game. Reddit is buzzing with devs calling OpenAI’s Codex “legit impressive” and “shockingly good.” With recent upgrades, Codex just edged out Anthropic’s Claude on GitHub merges, hitting a 74% success rate versus Claude’s 73.7%. Sure, Claude still crushes Codex in popularity, pulling 5m weekly downloads compared to Codex’s modest 190k. But devs chase performance, and Codex finally found its mojo. Anthropic’s coding monopoly might want to watch its back.

🦮 California just put AI companions on a leash. It’s now the first state requiring strict safety rules for chatbot companies after tragic incidents involving teens and AI crossing ethical lines. Companion-chat apps like Replika and Character AI, plus big players like Meta and OpenAI, face serious legal heat if their bots misbehave. Age verification, suicide-prevention measures, and transparency warnings are now mandatory. Looks like even digital friends need parental oversight.

🪖 Anthropic is getting cozy with the Pentagon. They once promised no military or surveillance uses, but apparently that’s changed. Claude will soon advise cyber operations, analyze foreign intelligence, and generate war-game scenarios for troops. Anthropic will even let government users experiment freely with new military uses, a huge reversal for a startup built on responsible AI. Ethics were nice while they lasted.

☀️ Australia is ditching coal for renewables. Not because they suddenly care about koalas or coral reefs, but because renewables are cheaper and more reliable. Fixing ancient coal plants just doesn’t make economic sense. Clean energy already supplies 77% of Australia’s grid at peak, and rooftop solar alone hits 55% on sunny days. Coal will vanish from 90% of the system by 2035, putting the country on track for 100% renewable power soon after. Who knew the free market would betray fossil fuels like this?

🤜🤛 Together with

Grow faster

Vanta helps you get compliant fast. It automates your compliance process so you can speed up your sales cycle and build deeper trust. 

Why you’ll love it: Trusted by startups and enterprise, Vanta builds credibility from day one. It gets you audit-ready in days, not months. 

Vanta is hands-down the fastest and most trusted way to prove security. Used by startups like Replit and Cursor to turn security into sales. Join them and save $1,000 today. 

🌽 Small towns are fighting back against AI data centers – and losing. Rural communities across the US are saying no to massive AI infrastructure projects, worried they’ll drain local resources or wreck their neighborhoods. One Michigan township rejected a new data center, only to get sued into submission. Ohio towns are passing outright bans, and Missouri cities are hitting pause completely. So much for small-town charm beating billion-dollar budgets.

🚨 OpenAI is using cops to silence critics. At least, that’s what Encode AI advocate Nathan Calvin says happened when a sheriff showed up demanding his private messages. Calvin claims OpenAI subpoenaed conversations he’d had with lawmakers, students, and even former OpenAI staff. OpenAI shrugs it off as part of a legal spat, accusing Elon Musk of secretly bankrolling its critics. Calvin calls it blatant intimidation aimed at scaring supporters of AI laws like California’s SB 53. Peter Thiel warned that AI regulation is basically the antichrist’s work, and OpenAI seems pretty eager to prove him right.

🚀 Warp drive used to be Star Trek fantasy. Now physicists want to build one for real. Recent breakthroughs show bending spacetime might be possible without breaking Einstein’s rules. The energy required is still mindbogglingly huge, but physicists have simplified the math, removing some of the biggest obstacles. “Warp drive” papers are now landing in legitimate physics journals, and serious scientists have stopped laughing. Boldly going where no one has gone before just got slightly less ridiculous.

fun stats

10 gigawatts

the latest OpenAI chip deal to power data centers, this time with Broadcom  Last week it was 6 GW with AMD, now there’s talk of a $25B data center project in Argentina

Over the past 3 weeks, Open AI has made deals for about 33 GW of compute. It’s currently operating on just 2 GW.  

[x]

🥷 Safe & secure: China’s Unitree G1 real time kungfu kid

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