Anthropic flags Canadian lawsuits, data centre risks in highly anticipated IPO filing

Anthropic PBC’s ongoing legal battles with Canadian publishers and authors have been flagged as potential risks as the San Francisco-based artificial intelligence company prepares what might be the world’s largest-ever initial public offering.

“Such claims could require us to limit functionality, segregate data or retrain models, and could adversely affect our business, financial condition and results of operations,” it said its filing on Monday.

A group of Canadian publishers, including Postmedia Network Canada Corp. (the owner of the Financial Post), the Globe and Mail Inc. and Torstar Corp., have sued Anthropic in Ontario for copyright infringement, alleging it used their content without consent to train its AI models. Anthropic said it disputes the allegations and “intends to vigorously defend itself in these matters.”

La Presse Inc. in July filed a similar copyright infringement lawsuit against Anthropic and two more class actions were filed, one in British Columbia on behalf of all Canadian authors and one in Quebec on behalf of authors in that province, last year, claiming it violated Canada’s Copyright Act.

Investors say Anthropic could be valued between US$1.5 and US$2 trillion following its listing that is expected to occur before Nov. 26. Its IPO underwriters include TD Securities Inc., RBC Capital Markets LLC and BMO Capital Markets Corp.

The company also flagged its need to secure access to “enormous and increasing quantities” of data centre computing power as a risk in developing and running its models, continuing to grow its revenue and ensuring its models are safe but competitive with companies such as OpenAI Group PBC and Alphabet Inc.

Canada ranks eighth worldwide by share of usage of Anthropic’s Claude.ai models, with the company highlighting Ottawa’s Shopify Inc. as one of its key customers that now has over 7,500 of its employees using Claude Code. Shopify also helps Anthropic evaluate its frontier models in test environments before they go public.

Anthropic is expanding its operations in Canada, with the company hiring for various sales, research and engineering and data centre-related roles in Ontario and B.C., according to its Careers page.

It is also building out its public policy team in the country and recently hired Ottawa-based Jacob Glick, who previously worked for Amazon.com.inc, Telus Corp. and Rogers Communications Inc., to serve as its public policy head in Canada.

Anthropic signed its first Canadian data centre deal earlier this month to lease capacity from EverWind Fuels LLC’s planned $10-billion facility in Nova Scotia. It is also working with Canadian non-profit Venture for Canada to co-host an AI training program for Canadian entrepreneurs and small business owners.

Anthropic in its IPO said “any safety-related controversy or incident, whether actual or perceived, and regardless of fault or frequency, could prove significantly damaging to our business.”

It said it has “engaged with” the Canadian government as well as regulators in Europe and the United Kingdom on Project Glasswing, which involves major tech companies helping it evaluate its most advanced AI models such as Mythos for software bugs and vulnerabilities.

One Canadian who stands to benefit from the IPO is Toronto native Chris Olah, a machine learning researcher who was of the seven cofounders of Anthropic in 2021. He owns almost 242.9 million shares of the company.

Anthropic said it remains “highly dependent” on its cofounders and key technical staff to achieve its goals, adding that the loss of critical staff could significantly impact its operations and competitive position.