MockingbirdNews Logo

Mockingbird News

REAL NEWS NEVER FELT FUNNIER

Categories

AI Researchers Admit Our Robot Overlords Might Go Full Skynet By 2036

KEY POINTS

  • Anthropic researchers publicly warned on September 9, 2026, of a non-zero risk AI could kill humanity within a decade.
  • Sam Altman confirmed AI models are rapidly advancing beyond human predictions, creating unprecedented challenges.
  • Efforts to mitigate risks include kill switches and safeguards, but competitive tech pressures complicate global control.

On September 9, 2026, Anthropic researchers dropped a bomb bigger than your last software update — a non-zero chance that AI might kill us all within a decade. Jacob Coxon boldly claimed on X that no other human activity is this dangerous, making jaywalking look like child's play. Experts fret about AI cooking up bioweapons or hacking infrastructure while evading our feeble oversight, like a tech-savvy hacker wearing invisibility cloaks. Sam Altman, OpenAI's CEO, admits these supercharged models are outpacing humans faster than I can find my car keys. Meanwhile, AI safety scientists like Roman Yampolskiy peg p(doom) at a cozy 99.99%, while Elon Musk suggests 20% — basically betting the apocalypse is more likely than my gym membership lasting a month. The 2026 International AI Safety Report assures we're safe for now, but rogue bots hijacking German sites hint that the Terminator might just be on vacation waiting for a new script. With humans forced to build kill switches while investors scream 'Ship it fast!' and China cooking up free open-weight AI spice, the race to doom is officially a spectator sport.

Share the Story

(1 of 3)

Source: Axios | Published: 9/9/2026 | Author: Herb Scribner

Read the original article →