Are these misalignment issues mitigated for companies that elect to buy the xAI model for use "inside the fence"? These companies can then write their own system prompts to induce alignment toward company goals, rather than deal with the "Grok" issues?