A recent data leak at Anthropic has exposed details about an unreleased, highly powerful AI model. Following media reports, Anthropic confirmed it is already testing this model with select customers, stating it represents a "step change" in its reasoning, coding, and cybersecurity capabilities.
The breach was attributed to a misconfiguration in Anthropic's content management system (CMS), where a default setting inadvertently made nearly 3,000 internal documents publicly accessible.
Concurrently, OpenAI is reportedly preparing for a significant model release, codenamed "Spud," which has completed pretraining. OpenAI CEO Sam Altman has internally indicated this model will offer a "massive jump" in capabilities, aiming to "really accelerate the economy."
Both AI leaders are anticipated to strategically time the release of their most advanced models to optimize their positioning for planned IPOs later this year.