Anthropic released its latest model Fable on Tuesday, billing it as a public and limited version of its powerful and much-hyped cybersecurity model Mythos.
But not everyone is happy with the restrictions, and a number of cybersecurity researchers and professionals have aired complaints online.
“[Fable] rejects any request that could be tangentially cyber related. Even innocuous tasks like reading a blog post,” said Valentina “Chompie” Palmiotti, a well-known security researcher who works at IBM X-Force.
When a prompt triggers its guardrails, Fable pauses the chat and says that its “safety measures flagged this message for cybersecurity or biology topics.”
The guardrails were put in place to limit the risk that Fable could be used to develop malware or compromise software — a longstanding concern within Anthropic. The restrictions on biology come from a similar concern around developing biological weapons.
When the AI giant released Mythos in April, it restricted the model to a limited number of companies and organizations in what it called Project Glasswing, an effort to deploy the model to secure critical software and infrastructure. Last week, Anthropic expanded access to Mythos to hundreds of organizations in 15 countries.
But despite the good intentions, many cybersecurity experts are still put off by the haphazard nature of the restrictions. Matt Suiche, a cybersecurity veteran, told TechCrunch that “if you ask it to write secure code, it assumes it is cybersecurity related work instead of software engineering best practices, and you get downgraded.” Fable is programmed to fall back to Claude Opus 4.8 if it hits a guardrail. “It seems to be keyword based, so anything in the lexical field of ‘cybersecurity’ triggers the guardrails.”
“But it is understandable as we are still in the early days and they are still adapting their guardrails. I am sure they are going to evolve over time as Anthropic and other frontier model companies will collaborate more with the current new generation of cybersecurity companies,” said Suiche, who is a member of the technical staff at Tolmo, an AI cybersecurity startup. “It’s better to catch more people than not enough when you do such a release and to relax the guardrails over time.”
Another researcher griped on X that “even asking for a code review” triggers Fable’s guardrails.
Anthropic did not immediately respond to a request for comment.
Apart from guardrails inside its models, Anthropic requires cybersecurity professionals to apply to the Cyber Verification Program. If they get approved, the applicants have fewer limitations on using Claude for cybersecurity work. OpenAI has a similar program called Trusted Access for Cyber.
Why enterprise AI will be a major focus at VivaTech 2026
TechCrunch is partnering with VivaTech 2026 to highlight the technologies, founders, and ideas driving the next wave of innovation. As part of the collaboration, TechCrunch and VivaTech will spotlight emerging startups through the VivaTech Innovation of the Year competition. The winner will earn a chance to pitch live in Paris and secure a place in Startup Battlefield 200 ahead of TechCrunch Disrupt 2026, taking place in San Francisco from October 13-15.
For anyone tracking the future of enterprise AI, VivaTech 2026 offers a front-row seat to some of the industry’s most important conversations. Register now to hear from the leaders building the next generation of AI infrastructure, applications, and operational systems.
Europe’s enterprise AI ecosystem is becoming impossible to ignore
For the past several years, the global AI race has largely been defined by foundation models, chatbot launches, and the battle for consumer attention. But beneath that public competition, another ecosystem has been gaining momentum — one centered on enterprise infrastructure, operational systems, and industrial AI.
While Silicon Valley continues pushing aggressively into large language models and consumer-facing AI products, many European companies are focused on applying AI to complex systems already embedded into everyday life: Manufacturing. Logistics. Healthcare. Cybersecurity. Energy infrastructure.
These industries are quickly becoming some of the most important battlegrounds in the AI economy. They also require far more than powerful models alone. That’s where Europe believes it may have an advantage.
Deploying AI inside large organizations introduces a different set of challenges altogether: governance, compliance, security, operational reliability, and long-term integration. In many ways, the industry is now confronting the realities of moving AI from experimentation to production at scale.
That shift will loom large at VivaTech 2026, which has increasingly become a showcase for Europe’s growing enterprise AI ambitions.
The AI industry’s next challenge
For many enterprises, the first wave of AI adoption was relatively experimental. Companies rushed to test copilots, automate workflows, and explore generative AI use cases across their organizations. But as the technology matures, the conversation is becoming significantly more complicated.
Now comes the hard part: Enterprises are confronting questions around governance, compliance, infrastructure, and security that many companies barely considered during the first wave of AI experimentation.
Increasingly, startups are being judged less on novelty and more on whether they can integrate into existing enterprise environments, navigate regulatory complexity, and deliver measurable operational value. Investors are starting to prioritize infrastructure, deployment, and measurable outcomes over pure experimentation.
Push the conversation forward at VivaTech 2026
At VivaTech 2026, those realities are expected to shape many of the conversations happening across the event floor.
Europe will argue that the next phase of the AI race may be won not just by building models, but also by deploying them effectively at scale. Join the discussion in Paris and see how founders, investors, and enterprise leaders are approaching AI’s transition from experimentation to production.
