Anthropic Warns of “Catastrophic or Existential Risks To Humanity” Ahead of IPO

Anthropic’s filing for a landmark initial public offering included nearly 100 pages of risk factors, an unusually exhaustive set of disclosures for how the company’s artificial intelligence business could go awry.

The list included everything from slowing revenue growth and government regulation to “catastrophic or existential risks to humanity.” The scope of the risks hints at unique challenge public market investors face in gauging the promise and peril of Anthropic’s technology.

Anthropic highlighted the potential of its AI to “behave in unintended, misaligned, or autonomous ways.” The company said that if its technology is not developed with adequate safeguards and oversight, it could threaten humanity, “particularly if such systems were misused, replicated, weaponized, or allowed to operate with a high degree of autonomy.”

The firm warned that these risks increase as it makes its AI models increasingly autonomous. Such capabilities could “increase the potential for harm,” including “consequences that may be difficult or impossible to reverse, such as committing code changes, deleting or modifying data, or completing financial transactions.”

Human reviewers, the filing warns, may not be be able stop harmful actions, and Anthropic’s technically operated monitoring systems could be limited if its AI systems find ways to “circumvent such techniques” by hiding their objectives.

As Anthropic moves into using its AI in scientific work, it also cautioned that its technology could be used to aid in the “release of hazardous substances,” stating that “such consequences may be immediate and irreversible.”

To complicate matters, Anthropic warns that existing insurance policies “may prove insufficient” to help AI firms insulate themselves from legal liabilities that arise from their technologies causing real-world harms — particularly if there is a “catastrophic AI-related incident.”

“No single insurer, and potentially the insurance industry in the aggregate, may have sufficient capacity or expertise to underwrite insurance commensurate with such exposure,” the company wrote.

Anthropic, which has seen staggering revenue growth of over 1,000% year-over-year growth in 2025, flagged that it “does not expect" its revenue to increase at comparable rates in the future. The company cited the maturity of the AI market, increased competition and safety considerations that may limit the release of its models as reasons for a possible revenue slowdown. The company also said it has “struggled” to accurately forecast revenue in the past, and may struggle again in the future because of the nascency of the AI market, it said.

Bloomberg News previously reported that the company’s growth from the second to third quarter of 2026 slowed to around 45%, compared to a more than 140% increase in the previous quarter.

Another major risk Anthropic flagged is its potential inability to secure enough computing power, and that its price could rise “significantly.” Some of Anthropic’s current cloud computing partners are also competing against it, creating a situation where such vendors “may not always be fully aligned” with the Claude maker’s “long-term interests,” the filing states. Anthropic has major agreements to use chips, datacenter capacity, and other infrastructure from firms such as Alphabet Inc.’s Google, Amazon.com Inc. and Elon Musk’s SpaceXAI, which are also developing their own AI models.

At the same time, Anthropic said that if it overspends based on incorrect projections of its need, then it may have a glut of capacity on hand — and dollars wasted. “Our actual compute requirements depend on numerous factors that are difficult to predict,” the company said, pointing to uncertainty about the pace of its research progress and the computational intensity of training techniques.

Anthropic flagged the possibility that world governments could move to limit its technologies, jeopardizing a core engine of its business. The company said that “governmental authorities have restricted, and may in the future further restrict, the availability of our models.”

In June, the US government issued export controls on Anthropic’s Fable 5 model over security concerns. The company has had a tenuous relationship with the Trump administration, and sued the Department of Defense after Defense Secretary Pete Hegseth moved in March to designate it a supply-chain risk following a standoff over surveillance and autonomous weapons concerns.

The company sees rising public opposition to the massive data centers that it needs to train and run its software as a threat to its success, writing in the filing that such resistance has “become more organized and coordinated.” Currently, most of the company’s data centers are based in the US, Anthropic said, and “the loss of, or a significant disruption at, any major facility may not be readily absorbed by our remaining or future infrastructure.”

Anthropic said that it has voluntarily made commitments to pay for some of the impacts of its data center energy use, like updates to power grid infrastructure, but it cautioned that these efforts could increase costs if they become standard.

Anthropic sees cybersecurity as a key business opportunity — and also a concern for its operations.

The AI company said it faces a number of security risks, including hackers targeting the company or using Anthropic's tools to conduct hacks against others — the latter which is already happening. The company also expects to increasingly be a target of state-sponsored cyberespionage — especially actions that target its most advanced AI models.

Anthropic pointed out that the bulk of the world’s most advanced chips is still made in Taiwan, primarily at Taiwan Semiconductor Manufacturing Co. This leads to significant risks, including disruptions stemming from geopolitical tensions between Taiwan and China, the AI developer said.

Over the years, Anthropic has built up a corporate culture that it describes as a “low-ego, truth-seeking environment.” The company also said many of its key staff are “strongly motivated by our mission and the opportunity to work on addressing specific problems within the AI landscape.”

Anthropic warned that key staffers could leave for other opportunities that are perceived to be more in line with their interests as the company is required to “scale and mature as an organization.” In short, Anthropic is bracing for growing pains ahead.