Share Facebook Twitter LinkedIn Pinterest Email Anthropic investigates elevated errors across Claude AI models