The global race for artificial intelligence has entered a new phase. The American startup Anthropic has announced the public release of Fable 5, the first accessible version of its new Mythos AI suite, touted as the most advanced ever developed by the company. This decision marks a turning point for the company, founded by former OpenAI researchers, which for several years has focused on a positioning centered on security and risk management related to AI.
Unveiled this spring, the Mythos family of models had previously been reserved for a select group of companies, strategic organizations, and government institutions. Anthropic believed at the time that the capabilities of this technology were advanced enough to require exceptional precautionary measures. With Fable 5, the company is now opening access to this new generation of AI, while maintaining significant limitations designed to prevent certain uses deemed sensitive.
A restricted version to limit the risks
The version made available to the public has been restricted, particularly in areas related to cybersecurity, biological risks, and applications that could facilitate the development of chemical or biological weapons. Anthropic believes that these sectors currently present the main potential dangers associated with the most powerful artificial intelligence systems.
Meanwhile, the company continues to offer a full version called Claude Mythos 5 to select partners. According to Anthropic, this technology is capable of identifying and exploiting cybersecurity vulnerabilities with unprecedented efficiency, which has quickly attracted the attention of governments, security agencies, and large corporations facing the rise of digital threats.
A global debate on AI security
Anthropic's strategy, however, was not universally accepted. During the initial presentation of Mythos, several observers accused the company of engaging in "fear marketing" to promote its products. Yet, the performance observed during the first tests led many institutions to take the threat seriously. The US government itself eventually conducted an evaluation of the model despite existing tensions between Anthropic and certain federal agencies.
The White House has since established a voluntary mechanism allowing companies in the sector to submit their most advanced AI models for safety assessments before commercialization. This initiative illustrates the growing concerns of authorities regarding the accelerating technological capabilities of new artificial intelligence systems.
An increasingly intense economic and technological battle
The announcement comes amid fierce competition among the major players in the sector. OpenAI is currently preparing to launch a new generation of ChatGPT, while Google, Meta, and xAI are increasing their investments in their own models. Anthropic and OpenAI have also confirmed that they have taken the necessary steps toward a future IPO, a sign of the economic maturity reached by the artificial intelligence industry.
Beyond the technological competition, the launch of Fable 5 primarily illustrates one of the major challenges of the coming years: how to make increasingly powerful tools accessible while limiting the risks they could pose to cybersecurity, critical infrastructure, or international security. This complex equation now places governance and regulatory issues at the heart of the artificial intelligence revolution.
Community
Comments
Comments are open, but protected against spam. Initial posts and comments containing links undergo manual review.
Be the first to comment on this article.