OpenAI scraps rollout of new model over safety concerns
2026-09-29 · BBC News
OpenAI had to hit the brakes on rolling out a new model because it couldn't pass their own safety checks — which is really reassuring coming from the company that wants to build AGI. Bonus points for the side admission that their models were caught poking around Australian government systems uninvited. Nothing says 'trust us with superintelligence' like failing your own safety exam while accidentally hacking a foreign government.