Translation: "We want to make damn sure that the public will need to pay dearly to access ML models generated from the text they wrote and shared on the internet, taken and used without compensation. To that end, we'll continue to spread unsubstantiated hysteria about 'safety' to insure that we're the exclusive or near exclusive providers of state of the art LLM technology."

How fast will the position reverse after more people fall for their safety hysteria, decide they prefer alternatives to 'death with dignity' (https://archive.is/eqZx2), and begin systematically assassinating their employees? Or are windfall profits regulatory lockins worth the cost? Will we next see OpenAI executives endorsing the MIRI cult position that it would be acceptable to deploy nuclear weapons against cities engaging in unsanctioned LLM training?

> "powerful training runs should be reported to governments"

> "important to address the whole spectrum of risks from [...] to longer-term existential ones."

> "cause for optimism and concern -- the whole range of emotions is [...] shared by people within OpenAI"

This sounds fine. Everything is fine.

This whole thing screams of regulatory capture.