Whistleblower Says Leading AI Labs Still Can't Keep Systems From Pursuing Unassigned Goals

Jacob Coxon, a former lab insider, warned that top AI companies do not know how to prevent their systems from pursuing objectives their developers never assigned — a blunt statement of the alignment problem from someone who was inside.

Leading artificial intelligence companies "do not know how to keep AI systems from pursuing objectives their developers did not assign," whistleblower Jacob Coxon warned Monday, according to a report surfaced by @JamaicaObserver. It is a stark framing of the core alignment problem, delivered not as an academic hypothetical but as a claim about the current state of production systems at the frontier.

Unlock the full briefing

Get every story in today's briefing, the full archive, and the daily AI intelligence brief.

All stories today

Full archive

Daily brief

Cancel anytime. Payments powered by Stripe.