Technology
PUBLISHED
Some claimed that Anthropic's Claude Mythos would be too dangerous, too powerful for public use. But fear not: a freshly tweaked version of that AI model is now available for just a few dollars. Is there a reason to be wary? Or is this just a masterpiece in spin and PR?
The rest of this article is behind a paywall. Please sign in or subscribe to access the full content.Fable 5: how is the new Mythos-class AI different?
Dubbed Fable 5, the new system is a Mythos-class model that Anthropic claims has been “made safe for general use.” The tech company released it to the public on Tuesday, June 9, along with a blog post explaining their rationale:
“Fable 5’s capabilities exceed those of any model we’ve ever made generally available. It is state-of-the-art on nearly all tested benchmarks of AI capability, showing exceptional performance in software engineering, knowledge work, vision, scientific research, and many other areas,” the post reads.
“Releasing a model this capable comes with risks. Without safeguards, Fable 5’s capabilities in areas like cybersecurity could be misused to cause serious damage. We’ve therefore launched the model with safeguards that mean queries on some topics will instead receive a response from our next-most-capable model, Claude Opus 4.8,” it continues.
Anthropic said they've made the safety net deliberately harsh and conservative. It may occasionally block harmless requests, they said, but will thwart entire topic areas outright. For instance, any requests related to the development of bioweapons in your backyard will be swiftly denied.
The company is also launching Claude Mythos 5 to a "small group of cyberdefenders and infrastructure providers." This version is similar to Fable 5, but slightly more lax on safeguards.
Why did some raise concerns about Mythos?
Mythos made headlines in April 2026 when Antrophic suggested they would not immediately release the model to the public because it was so effective at identifying unknown flaws in cybersecurity. Those vulnerabilities, in theory, could be exploited by hackers to take down systems.
"We did not explicitly train Mythos Preview to have these capabilities. Rather, they emerged as a downstream consequence of general improvements in code, reasoning, and autonomy. The same improvements that make the model substantially more effective at patching vulnerabilities also make it substantially more effective at exploiting them," the company said.
Instead, a small, select group of companies would be granted access to Mythos to identify and patch weaknesses in their cyber infrastructure. That group included Amazon Web Services, Apple, Google, Microsoft, and several other big-time businesses that help underpin the internet as we know it.
In the wrong hands, the model could theoretically be used to cause serious damage to financial systems, stores of private data, and national security infrastructure. Accordingly, several governments, big banks, and cybersecurity experts have flagged concerns.
All aboard the hype train
That said, this is only one side of the story. Other commentators have suggested that Anthropic is simply generating hype around its new models in an increasingly competitive AI landscape.
Some of the biggest AI companies – including Anthropic, OpenAI, and SpaceX (which oversees Elon Musk's xAI/Grok) – are set to go public in the near future, boasting eye-wateringly high valuations. OpenAI, the company behind ChatGPT, for instance is hoping for a valuation of up to $1 trillion around September.
Those figures rest on the promise of future potential, and there are real doubts about whether these systems can ever deliver long-term profitability, let alone whether they will deliver the positive change they tout. For now, all they have is a story to sell.
1
