The maker of ChatGPT has advertised a $555,000-a-year vacancy with a daunting job description that would cause Superman to take a sharp intake of breath.<\/p>\n
In what may be close to the impossible job, the \u201chead of preparedness\u201d at OpenAI<\/a> will be directly responsible for defending against risks from ever more powerful AIs to human mental health, cybersecurity and biological weapons.<\/p>\n That is before the successful candidate has to start worrying about the possibility that AIs may soon begin training themselves amid fears from some experts they could \u201cturn against us\u201d.<\/p>\n \u201cThis will be a stressful job, and you\u2019ll jump into the deep end pretty much immediately,\u201d said Sam Altman<\/a>, the chief executive of the San Francisco-based organisation, as he launched the hunt to fill \u201ca critical role\u201d to \u201chelp the world\u201d.<\/p>\n The successful candidate will be responsible for evaluating and mitigating emerging threats and \u201ctracking and preparing for frontier capabilities that create new risks of severe harm\u201d. Some previous executives in the post have lasted only for short periods.<\/p>\n The opening comes against a backbeat of warnings from inside the AI industry about the risks of the increasingly capable technology. On Monday, Mustafa Suleyman, the chief executive of Microsoft AI, told BBC Radio 4\u2019s Today programme: \u201cI honestly think that if you\u2019re not a little bit afraid at this moment, then you\u2019re not paying attention.\u201d<\/p>\n Demis Hassabis, the Nobel prize-winning co-founder of Google DeepMind, this month warned of risks<\/a> that included AIs going \u201coff the rails in some way that harms humanity\u201d.<\/p>\n Amid resistance from Donald Trump\u2019s White House, there is little regulation of AI at national or international level. Yoshua Bengio, a computer scientist known as one of the \u201cgodfathers of AI\u201d, said recently: \u201cA sandwich has more regulation than AI.\u201d The result is that AI companies are largely regulating themselves.<\/p>\n Altman said on X as he launched the job search: \u201cWe have a strong foundation of measuring growing capabilities, but we are entering a world where we need more nuanced understanding and measurement of how those capabilities could be abused, and how we can limit those downsides both in our products and in the world, in a way that lets us all enjoy the tremendous benefits. These questions are hard and there is little precedent.\u201d<\/p>\n One user responded sardonically: \u201cSounds pretty chill, is there vacation included?\u201d<\/p>\n What is included is an unspecified slice of equity in OpenAI, a company that has been valued at $500bn.<\/p>\n Last month, the rival company Anthropic reported<\/a> the first AI-enabled cyber-attacks in which artificial intelligence acted largely autonomously under the supervision of suspected Chinese state actors to successfully hack and access targets\u2019 internal data. This month, OpenAI said<\/a> its latest model was almost three times better at hacking than three months earlier and said \u201cwe expect that upcoming AI models will continue on this trajectory\u201d.<\/p>\n