Anthropic has released Claude Fable 5, its most powerful AI model to date, incorporating a version of capabilities previously known as "Mythos"—features the company had itself flagged as potentially too dangerous to release publicly just months earlier.
According to the Indian Express, Mythos was considered too risky for public deployment as recently as April. The outlet poses the central question now circling the AI industry: if the capabilities were dangerous then, what changed by June? The answer, based on available reporting, appears to be a combination of new safeguards and a deliberate decision to split access to the model.
According to SecurityBrief Australia, Anthropic has "split" Claude Fable 5 amid cyber risk fears—suggesting the company is offering tiered or restricted versions of the model rather than a single open release, with more powerful capabilities gated behind additional controls or enterprise agreements.
The launch is already rippling through financial markets. ERP Today reports that Fable 5's release has renewed pressure on enterprise software stocks and valuations, as businesses reassess how advanced AI models might displace or reshape traditional software categories.
On pricing, questions are mounting. A report via MSN and Euronews asks whether Fable 5 is worth its cost, noting that OpenAI may soon offer a cheaper alternative—framing the release as the opening move in a fresh round of competitive pressure between the two leading AI labs.
The Fable 5 launch crystallizes a tension at the heart of frontier AI development: how to release genuinely powerful capabilities responsibly, and who gets to decide when "dangerous" becomes "manageable"—a question with stakes far beyond any single product launch.