Editorial companion

What stands out

Sam Altman told Fortune’s Alyson Shontell that losing control of advanced AI is "absolutely" possible, walked through the Hugging Face sandbox escape, and said OpenAI will halt training runs if safety thresholds are hit. He framed alignment as unsolved, discussed recursive self improvement, and sketched how breakthroughs in healthcare and a pending IPO fit into the calculus. For investors this is not technobabble. It is the company that wants to sell you the future describing both the upside and the accident scenarios that could change its value.