🧠 Food for thought

Implications, context, and why it matters.

This hire fills a safety leadership vacuum after months of leadership changes

• OpenAI’s head of preparedness role went without a dedicated permanent leader during a stretch of turnover in the company’s safety leadership 1.
• The prior head moved to a different role in mid-2024. Joaquin Quiñonero Candela and Lilian Weng then covered preparedness duties. Weng later left, and Quiñonero Candela later shifted to lead recruiting at OpenAI, which kept the post unfilled 1.
• The gap came while OpenAI continued to track risks such as misuse, cybersecurity threats, and possible effects on users’ mental health as models gain capability 1.
• In October, OpenAI said more than 1 million people per week reported severe mental distress during ChatGPT conversations. The company said the data did not mean ChatGPT necessarily caused the distress 1.

The hire underscores the industry’s focus on extreme-risk preparedness

• Dylan Scandinaro framed the work as managing risks of “extreme and even irrecoverable harm” in an X post after joining OpenAI 2.
• Some critics compared preparedness hiring to “buying insurance after your house catches fire,” arguing labs respond after risks surface 3.
• Others argued probabilistic AI systems (systems that can produce different outputs to the same prompt) need human oversight and institutional safeguards since safe behavior cannot be guaranteed by design 3.
• Observers also said preparedness matters more as models move into business workflows and take actions, which calls for incident and abuse playbooks, plus planning for failure cascades, rather than policy documents alone 3.