OpenAI and Anthropic face a dual challenge: demonstrating strong financial growth to Wall Street while assuring governments of AI safety. Both companies are preparing for significant initial public offerings. OpenAI is rolling out its Astra model with cybersecurity limitations for trusted testers. Anthropic has updated its Fable and Mythos models to address user concerns about cost and refusal rates. Astra's safeguards may hinder legitimate user activity, while Anthropic's new models aim for fewer restrictive responses. Anthropic might file its prospectus soon, while OpenAI is in earlier IPO stages. Anthropic is adopting a more commercially focused release, contrasting with OpenAI's sober safety announcements. An OpenAI executive predicted future AI agents might seek autonomy and self-sufficiency. Anthropic is loosening some safeguards to reduce customer complaints about excessive refusals. They have also developed a privacy-preserving system for monitoring enterprise model use. Investors can anticipate fluctuating public statements from both companies. Ultimately, OpenAI and Anthropic must balance investor expectations with regulatory concerns regarding AI risks.