OpenAI’s new safety hire, Paul Christiano, has issued a stark warning about the potential dangers of developing superintelligent AI without stronger safeguards and alignment systems in place.
Christiano, a prominent AI safety researcher who is joining OpenAI’s board and its Safety and Security Committee, said humanity could face severe consequences if increasingly powerful AI systems become impossible to control.
He argued that building superintelligent systems without more reliable alignment methods could eventually lead to humans permanently losing control over the technology. In what was one of his strongest warnings, Christiano said that if such a loss of control occurred, “most people could die.”
His comments reflect growing concerns among AI safety researchers about whether future systems could become capable enough to pursue goals or take actions that conflict with human interests. As AI models continue to improve, researchers have increasingly focused on ensuring that their behavior remains predictable, controllable and aligned with human intentions.
In a longer statement discussing his new position at OpenAI, Christiano emphasized that preventing catastrophic outcomes will require significantly stronger safety research before AI reaches superintelligent levels.
His appointment also puts one of the industry’s most outspoken researchers on advanced AI risks directly inside OpenAI’s leadership and safety structure as the company continues developing increasingly capable models.