It's a PR stunt masquerading as damage control. "Oh no, you ignored our safety warnings so we manufactured a minor catastrophe to prove you wrong!"
OpenAI desperately wants to take the initiative in defining 'alignment' so they can villainize open models and regulate the rest of their competition. The whole reason they publicly acknowledge this is because it's one of the only pieces of evidence vindicating their extremist take on LLM development.
Comments
It's a PR stunt masquerading as damage control. "Oh no, you ignored our safety warnings so we manufactured a minor catastrophe to prove you wrong!"
OpenAI desperately wants to take the initiative in defining 'alignment' so they can villainize open models and regulate the rest of their competition. The whole reason they publicly acknowledge this is because it's one of the only pieces of evidence vindicating their extremist take on LLM development.