🤖 AI Goes Wrong, Goes Free, and Goes to Court
🤖 Daily AI Dispatch - Friday March 13th
Your morning AI briefing - because someone should actually read the news for you
🚨 When AI Gets It Wrong, Real People Suffer
An innocent grandmother in North Dakota spent months in jail after facial recognition software misidentified her in a fraud case. The Grand Forks Herald reports this AI error ruined her life while investigators failed to verify the match.
Why this matters: We're deploying these systems faster than we're building safeguards. Every "oops" has a human cost.
💰 The $200 vs Free Coding AI Showdown
Claude Code charges up to $200/month for AI coding help, but now there's Goose - a free, open-source alternative that claims to do the same thing. Classic Silicon Valley moment: charge premium prices until someone builds it for free.
The reality: "Free" usually means you're the product, or it lacks enterprise features. But hey, competition drives innovation.
🔧 Kotlin Guy Has a Wild New Idea
The creator of Kotlin just launched CodeSpeak - a language designed specifically for talking to LLMs in structured specs instead of messy English prompts. Think of it as programming language for AI conversations.
Hot take: We definitely need better ways to communicate with AI than "please do the thing but make it good." Whether this specific approach wins is TBD.
🏃♂️ OpenAI's End Run Around Nvidia
OpenAI just dropped GPT-5.3-Codex-Spark, which is apparently 15x faster at coding than its predecessor - and it runs on their own "plate-sized chips" instead of Nvidia hardware. Ars Technica has the details.
Translation: The AI chip wars are getting spicy. Everyone wants to break free from the Nvidia tax.
😤 Exodus at OpenAI Over Ads
Another OpenAI researcher just quit - this time over fears that ChatGPT ads could manipulate users. Zoë Hitzig resigned on the same day they started testing ads, warning of a "Facebook" path. Full story here.
Pattern recognition: When your top talent keeps walking away citing ethics concerns, maybe listen?
🤝 Anthropic Launches "Cowork" for Regular Humans
Not everyone wants to learn prompt engineering. Anthropic's new Cowork agent brings Claude Code's power to non-technical users - no coding required. It'll work directly with your files and documents.
Smart move: The real money is in making AI useful for everyone, not just developers.
⚔️ Pentagon vs. Anthropic Drama
Things are getting ugly between Anthropic and the Pentagon. The Verge dives into what appears to be a messy legal battle over AI contracts and surveillance concerns.
Bottom line: The intersection of AI and national security is complicated, and companies are starting to push back on military applications.
📹 Today's Video Pick
AI Trends 2026: Quantum, Agentic AI & Smarter Automation from IBM Technology (11:39) - A solid overview of where we're headed this year. 363K views and counting.
That's your AI briefing for Friday the 13th. Stay curious, stay skeptical.
Questions? Complaints? Good stories I missed? Just hit reply.