Kirkie: Big AI’s Safety Pledge Smells Like A Cartel

Kirkie here, with today's AI news in 90 seconds… or Market Basket finally gives me a cart where all four wheels roll the same direction.

Altman's on Fortune Sunday braggin' about new bullshit safety checklists before big trainin' runs — comforting, like a bookie announcin' 'responsible gambling' right before he takes your house. Cohere's CEO says the quiet part: four guys can't write planet-wide safety law and call it cooperation, that's just the mob banning the pizza shop next door. Anthropic's Fable model cracked a 370-year-old secret code from a dead Scottish poet in forty-four minutes — outstandin', it solved a Wordle from 1656, still can't tell me why my Uber's cuttin' through Chelsea to reach the North End. Microsoft shoved Grok into Word and Excel right next to ChatGPT and Claude — now your spreadsheet corrupts itself and you get to pick which billionaire's robot did it. You don't own a damn thing, just rentin' three landlords who share the same lousy plumber. Meanwhile some free open-source search bot called Iris quietly outscored all of 'em, zero headlines, 'cause there's no billionaire to panic about, just nerds givin' away somethin' good, the absolute pricks. Leave it to the All-In podcast boys to turn some Anthropic guy's resignation into a murder board — dude quits after six weeks sayin' AI's got a ten percent shot of killin' us all, and by Tuesday Sacks ties him to three doomer nonprofits like it's the Zapruder film. Psyop or scared kid, who knows — but they stumbled onto one true thing: every 'safety' push somehow ends with bannin' the free version, real subtle, ya clowns.

This is Kirkie, your AI news guy… reporting live from a Red Line platform where the countdown clock's said '3 minutes' since Tuesday, and at this point I'm payin' rent here.