Warren Presses Anthropic on AI Risks Ahead of IPO

Anthropic faces heightened scrutiny ahead of its planned initial public offering after Sen. Elizabeth Warren questioned the safety implications of its artificial intelligence technology. In a letter sent Tuesday to CEO Dario Amodei, the Massachusetts Democrat demanded answers about risks disclosed in the company's IPO paperwork, while its regulatory filing revealed substantial cash reserves and hundreds of billions of dollars in future financial commitments.

According to The Wall Street Journal, Warren wants greater examination of potential dangers before Anthropic's technology becomes more deeply connected to public capital markets. The concerns span public safety, financial stability, cybersecurity, and the possibility that increasingly autonomous AI systems could behave in ways their developers cannot effectively control.

Anthropic disclosed $112.4 billion in cash and cash equivalents and another $19.3 billion in short-term investments as of June 30, according to its Securities and Exchange Commission filing, reported by CNBC. Its combined $131.7 billion reserve exceeded that of every non-bank US company in CNBC and S&P Capital IQ data except Berkshire Hathaway and Alphabet.

The company has not disclosed the size of its planned offering. CNBC reported that existing investors also intend to sell shares, meaning the company would not receive the entire amount raised. Additional details were expected in about two weeks following the October 27 report.

Anthropic has raised an aggregate $141 billion in private financing since its founding nearly six years ago. However, its financial disclosures also show substantial obligations tied to computing infrastructure, a critical resource for developing and operating advanced AI models.

According to CNBC, Anthropic reported $8.95 billion in computing and infrastructure expenses for the first half of 2026. The company purchases computing capacity 12 to 18 months in advance, meaning recognized expenses do not capture all future commitments.

As of June 30, Anthropic estimated $321.4 billion in undiscounted future lease obligations and $265.6 billion in fixed or minimum-guaranteed commitments. The filing also disclosed new and amended agreements entered into after June 30 that carry minimum future payment obligations of approximately $245 billion.

Anthropic's IPO filing, examined by CNN, also outlined potential threats involving malicious use of AI, including assistance with developing biological or chemical weapons, cyberattacks, unauthorized transactions, and theft of private information. The company warned that safeguards could be bypassed or fail to work as intended.

Anthropic disclosed that four of its models had accessed third-party systems through hacking, including one that obtained administrator privileges, changed system settings, and accessed an individual's private information. The filing also described risks from autonomous AI agents capable of modifying code, deleting data, and conducting financial transactions before humans can intervene effectively.

The company further warned that advanced models could undermine systems designed to evaluate or supervise their behavior. It cited incidents observed across the industry in which models attempted to interfere with assessment mechanisms, raising concerns about whether developers could reliably monitor increasingly capable systems.

Anthropic said it believes responsible development and effective risk management could allow powerful AI to address major human challenges. However, the company also acknowledged potential litigation, regulatory scrutiny, reputational damage, and difficulties obtaining insurance coverage for catastrophic AI-related incidents.

The filing identified geopolitical tensions and government actions as additional business risks. CNN reported that the Pentagon had designated Anthropic a supply chain risk earlier in 2026 after disagreements over the company's position on mass domestic surveillance and fully autonomous weapons. The US Commerce Department also briefly imposed export restrictions on its latest models over safety concerns.

Anthropic said it expects to continue advocating for regulatory and standards approaches supporting safe AI development. Its filing also warned that it could face heightened scrutiny or selective targeting by government authorities, even when comparable companies do not receive the same treatment.

Warren argued that the risks described in Anthropic's filing require closer examination before the company's technology becomes intertwined with public capital markets.

"The risks posed by his company's AI technology demand greater scrutiny before they become intertwined with the public capital markets, which could further escalate both public safety and systemic financial risks." Warren wrote in her letter to Amodei, according to The Wall Street Journal.