Anthropic warns safety rules drove customers to rival AI providers ahead of IPO

Anthropic’s S-1 filing for its upcoming Nasdaq Global Select Market IPO highlights a tension between its safety-monitoring policies and client demand for strict data privacy. Under Zero Data Retention agreements, the firm retains only classifier scores, preventing offline analysis, human review, or historical data checks, which has led some customers to migrate to rival AI platforms. The company also acknowledges that delaying model rollouts and limiting functionality for safety testing may have negatively impacted short-term revenue, citing the Mythos model as an example.

Financially, Anthropic posted $16.3 billion in revenue but recorded an $80.7 billion net loss and a $693 million operating loss for the six months through June 2026, an improvement from the prior year’s $2.1 billion operating loss. The filing does not identify the departing customers, competing firms, or the exact revenue loss. Anthropic warns that its safety-first approach could further affect adoption, market perception, and valuation, and it cannot guarantee that the long-term benefits will materialise as expected.