Anthropic has launched Claude Mythos 5, a brand new AI model, on 9 June 2026, even as critics question whether the company has properly assessed the risks involved. This stunt adds to the growing anxiety that industry leaders are prioritising market speed over user safety.
What Exactly Was Unleashed?
Claude Mythos 5 is Anthropic’s newest incarnation of the Claude series, announced alongside the companion release of Claude Fable 5. The company describes the tool as “safe” and suitable for public deployment, promising robust safety mitigations. The releases went live at 17:08 GMT on 9 June 2026, the same day a BBC report raised alarms about the company’s approach to risk.
Users can now access the model via Anthropic’s public API and playground. The company indicates that Mythos 5 has been audited internally and that safety layers have been enhanced compared to earlier iterations. Nonetheless, the term “safe” has been criticised as ambiguous and potentially misleading.
Who Is Behind the Release?
Anthropic, the San‑Francisco‑based AI research organisation, founded in 2020 by ex‑OpenAI employees, is the sole developer of Claude Mythos 5. The company’s CEO, Dario Amodei, has repeatedly touted the model’s safety features in press releases. The fourth‑generation Claude, known as “Fable,” was introduced a year earlier, and the new Mythos 5 builds on that architecture. No external partners were named in the launch, suggesting Anthropic is handling the rollout in-house.
The BBC, through its monitoring of AI ethics, is the only external source reporting on the launch, highlighting that the company moved forward despite risk concerns. The Guardian published a similar headline, echoing the BBC’s wary tone but emphasising the company’s claim that the model is safe.
What Did Anthropic Say?
In a terse press note, Anthropic said, “Claude Mythos 5 is a safe, publicly available model that incorporates the latest safety research.” The note added that the release follows an internal review process and aligns with the company’s stated commitment to responsible AI.
No additional commentary was provided from Amodei or other senior executives. The language remains deliberately vague, offering no concrete safety metrics or third‑party validation.
Why This Matters
Anthropic is one of the few companies that has claimed to focus on building truly safe large‑language models. Each new release is therefore met with high expectations—and high scrutiny. The risk of deploying models without rigorous safety checks could lead to misinformation, biased outputs, or unintended behaviour.
Public deployment also raises questions about regulatory oversight. In the UK, the AI Act is still under discussion, but the European Union’s approach to safety-critical AI differs markedly from the anecdotal safety claims made by private firms, leaving a regulatory grey zone for models like Mythos 5.
Reactions From the Public and Experts
Tech commentators have offered mixed reviews. Some praise Anthropic’s boldness and the availability of a new tool for developers. Others, however, echo the BBC’s scepticism, arguing that the term “safe” is marketing fluff unless backed by independent audit.
The Guardian’s coverage frames the release as an “interesting” development but does not mince words about the potential dangers of unverified AI systems entering the market.
What Feels Tense For Us?
For developers who rely on AI for creative or business tasks, the arrival of Claude Mythos 5 offers a tempting new resource. Yet the lack of transparency around its safety protocols is a red flag. Whether the model behaves as promised remains to be seen.
Given the speed at which AI firms are dropping new tools, this launch reminds us that the conversation about safety must keep pace with innovation. If companies continue to brand their products as safe without proof, users will only be left with a hoax of a new capability and a precarious future for trustworthy AI.