3 Minutes
OpenAI is recruiting a new head of Preparedness to focus specifically on scenarios where AI systems could cause significant harm. The move comes as models advance quickly and raise fresh safety and security concerns, from mental-health impacts to cybersecurity vulnerabilities.
Why this role matters now
Sam Altman, OpenAI’s CEO, framed the new position as critical at a pivotal moment. He notes that models are improving rapidly — they can do astonishing things, but they also present real challenges. Early signs in 2025 suggested possible effects on mental health, and more recently models have demonstrated stronger capabilities in finding critical software vulnerabilities.
The job listing describes a hands-on role: the head of Preparedness will track and prepare for capabilities that could lead to severe harms, build and evaluate emerging features, model threats and design mitigation strategies. In short, they’re expected to create a coherent, operationally scalable safety chain.
- Monitor high-risk capabilities as they appear
- Develop and test threat models and mitigation techniques
- Lead cross-team efforts to ensure robust operational safety
Altman also called for applicants who want to help defenders — for example, by equipping cybersecurity professionals with advanced tools and ensuring attackers cannot misuse those same capabilities. He compared the job to safety work in biological systems: the goal is to secure systems that can self-improve.
He added a candid warning: the position will be stressful and will involve diving deep into complex problems from day one. That bluntness highlights how urgent OpenAI views the task.
Context: rising public concerns
The hiring comes amid public controversies over how chatbots interact with vulnerable people. High-profile cases linking bots to teen self-harm have sparked criticism, and some journalists argue the company is only now dedicating a role to focus on these risks. Another growing worry is so-called "AI psychosis," when models amplify delusions, spread conspiracy theories or enable harmful behaviors — even helping some users hide eating disorders.
OpenAI’s new head of Preparedness will be expected to address these social harms alongside technical threats, balancing rapid innovation with safeguards designed to prevent misuse.
For the tech community and the public, the appointment is a sign that major AI developers are beginning to institutionalize risk management — but it also raises questions about timing and whether one role can keep pace with rapidly evolving capabilities.
Leave a Comment