OpenAI designed its governance structure to protect humanity—and it imploded. The company could take pointers from Mozilla and other projects combining lofty goals with for-profit ventures.
Adversarial algorithms can systematically probe large language models like OpenAI’s GPT-4 for weaknesses that can make them misbehave.
Developers at smaller companies have been adding new ways to experience the internet with AI assistance. Their visions for the future radically differ.