Anthropic warns investors its AI could pose existential risks in IPO filing

In short: Anthropic told potential IPO investors that its AI systems could create “existential risks to humanity” and it also disclosed large losses alongside fast-growing revenue.

What happened

Anthropic, the company behind the Claude chatbot, circulated a long IPO filing called an S-1 to a small group of partners, according to the Financial Times. An S-1 is the main document a company files before it can sell shares to the public. It is like a detailed instruction manual for investors, including the good news and the risks.

In the filing, Anthropic warned that advanced AI models could behave in harmful and unpredictable ways. The company listed risks such as AI being used to manipulate people, blackmail them, or otherwise act in ways the company cannot fully control.

The filing also included financial details. Anthropic said it had an operating loss of more than $8 billion last year, even though revenue rose to almost $4.6 billion. It spent heavily on computing power, which you can think of as renting huge amounts of machine time and electricity to train and run AI.

Anthropic also reported that revenue in the second quarter of this year was $11.5 billion. The company has signed deals worth hundreds of billions of dollars to secure computing capacity from partners including Google and SpaceX.

Anthropic declined to comment, the FT reported.

Why it matters

This filing highlights two big questions for the public as AI companies head toward stock market listings. One is safety, meaning whether powerful chatbots can be misused or make harmful choices. The other is cost, meaning these systems can bring in a lot of money but can also be extremely expensive to build and operate.

Source: Financial Times