“Pat Gelsinger considers recent announcements from OpenAI regarding erotica and from Anthropic regarding their inability to validate models to their own standards as 'not okay'.”