OpenAI disbanded its readiness team at the end of July, which assessed the risks of its most powerful artificial intelligence models. The Financial Times reported this, noting that her duties have now been distributed to other parts of the company.
The team examined scenarios in which advanced artificial intelligence could independently attack other people’s systems, use cyber tools, or enable biological threats. Existing departments will now be involved in assessing risks in these areas. OpenAI believes that the work itself is not disappearing anywhere, and integrating it directly into the development of models and products can make safety systems more effective.
However, the current decision appears particularly striking against the backdrop of previous reshuffles. Over the past two years, the company has shuttered its AGI Readiness and Mission Alignment divisions, which were also responsible for security and compliance for the OpenAI mission. In addition, chief operating officer Brad Lightcup and chief ethics officer Chloe Bakalar also left the company. Former preparedness chief Dylan Scandirano is now focused on the risks of self-improving artificial intelligence, even though the team he built no longer exists.
