On 10th September 2026, Andrew Grill, the Actionable Futurist®, joined Becky Anderson on CNN International’s Connect the World for an eleven-minute live interview on the AI safety story dominating headlines this week, former Anthropic researcher Jacob Coxon’s resignation, and his warning, in an interview with Anderson Cooper, that AI labs are “racing straight to self-improving superintelligence.”
Speaking from London, Andrew brought a practitioner’s view, drawn from hundreds of client engagements, on what the warning means beyond the lab: alert, not alarmed, and a wake-up call that boards can’t afford to treat as background noise.
What Andrew covered:
- Why the AI safety warnings deserve to be taken seriously, even if extinction within four years isn’t realistic
- The “capture the flag” detail from OpenAI’s own review, AI agents pursuing a task with no instinct to flag a problem to a human
- Why most boardrooms don’t yet know what their own AI agents can access, or who’s checking
- A workable model for regulation, borrowed from aviation and pharma: mandatory incident reporting and a tested kill switch
- Why AI won’t eliminate jobs wholesale, but will strip out the inefficient tasks inside them, “focus on what you love, automate the rest”
“It’s right to be alert, but not alarmed.” — Andrew Grill, CNN International, Connect the World
