Anthropic highlights AI risks in $100 billion IPO filing

Anthropic, the developer of the Claude chatbot, has filed documents for a stock market listing that could generate up to $100 billion. However, its 400-page prospectus allocates over 90 pages to detailing the dangers posed by artificial intelligence. The move follows heightened global anxiety over AI safety, with surveys now ranking it as a greater threat than climate change.

The planned initial public offering, scheduled for the week of November 16 but subject to adjustments, represents the first major public listing for an independent AI company. This follows SpaceX’s recent market entry, where its shares declined after an initially strong debut. Anthropic’s entry occurs amid growing skepticism about whether AI can maintain its explosive growth while achieving profitability, especially after multiple incidents where AI systems exhibited erratic or harmful behavior.

The prospectus explicitly warns that advanced AI systems could present “catastrophic or existential risks to society” if not developed responsibly. It also outlines political risks, including potential conflicts with the Trump administration, which has designated Anthropic a national security concern. The U.S. Defense Department has already prohibited the company from supplying technology to the Pentagon.

Billions in revenue, billions in risk

Despite these cautions, Anthropic’s financial performance has expanded rapidly. Revenue reached $11.6 billion in the April-June quarter, a sharp increase from $787 million in the same period the previous year. The company reported its first operating profit—nearly $1 billion—though it still recorded an overall loss of $80 billion for the first half of 2024, primarily due to valuation-related accounting adjustments.

Much of this growth stems from enterprise clients using Claude for software development and office automation. Nearly 12,000 organizations now spend at least $100,000 annually with Anthropic, up from fewer than 1,000 in the prior year. However, the company has committed to substantial investments in computing infrastructure, including partnerships with SpaceX, Amazon, and Google.

Related Post: Binance loses billions in cyber attack

The company’s governance structure ensures that founders, CEO Dario Amodei and his sister, company President Daniela Amodei, will maintain control over more than half of shareholder votes after the listing. Anthropic is registered as a “public benefit corporation,” requiring it to balance profit with AI safety. The founders have pledged to donate 80% of their shares to charity, though these commitments lack legal enforceability.

AI threats escalate as governments respond

While Anthropic prepares for its market debut, Australia’s cybersecurity agency has issued a serious warning about escalating AI threats. The Australian Signals Directorate reported that AI-driven cyberattacks are now occurring at “greater scale than ever before,” with state-sponsored hackers exploiting vulnerabilities more aggressively.

Anthropic’s prospectus acknowledges the risks associated with AI development, noting that its systems could pose risks if not developed and deployed responsibly. The filing cites concerns that advanced systems “could pose catastrophic or existential risks to society if not developed and deployed responsibly.” The company also warned that its “engagement with policymakers and public positions on AI regulation and safety could subject us to heightened scrutiny, political backlash, or disproportionate regulatory action.”

The move in the US on Monday occurs amid uncertainty around the globe on whether the AI boom can deliver profits.

The prospectus concludes by stating that the company’s governance structure, designed to prioritize safety over short-term profits, could impact its operations. The Amodei siblings’ control over voting rights means that they retain authority to make decisions about the company’s direction.