Anthropic Creates AI Model that Overcomes Humans in Cybersecurity and Promises to Debate Risks Openly

Anthropic reveals Mythos: Cybersecurity AI that requires debate

The artificial intelligence company Anthropic announced the creation of his new model, named Mythos. In a movement that combines technological advancement with caution, the company stated that it will expose the risks at the table' ("put the risks on the table") while deliberately limiting the initial implementation of the system. The announcement, as reported by the Digital Journal, highlights that Mythos has advanced capabilities in the field of cybersecurity, raising complex issues for the sector.

AI overcomes human capabilities in cybersecurity

Anthropic's CEO, Dario Amodei, described Mythos as a model that is beginning to overcome human capabilities'** in cyberspace-related tasks. This statement positions Mythos not only as a support tool, but as a potentially autonomous agent in critical areas such as vulnerability identification, incident response and threat analysis. The decision to restrict immediate access reflects the company's awareness that such skills, if misused, may pose significant risks.

Strategy to put the risks of AI on the table

The posture declared by Anthropic goes beyond the simple creation of technical guards. The company proposes an open debate on potential dangers associated with AI models with such specialized and powerful capabilities. This includes publicly discussing misuse scenarios, security failures and the impact that superior AI agents on cybersecurity can have on geopolitical balance and national security. The approach echoes concerns raised by other experts, such as Anthropic's initial investor, who has already warned about the need to educate ourselves about the risks of AI**.

IA implications in cybersecurity for companies and governments

The development of Mythos occurs at a time of intense acceleration of investments in AI. Anthropic itself received a massive investment of up to US$ 25 billion (/noticias/amazon-amplia-aposta-em-ia-com-investimento-de-ate-us-25-bilhoes-na-anthropic), demonstrating the strategic value attributed to the technology. The arrival of a model with capabilities described as superhuman in cybersecurity forces companies and governments to rethink their defenses. On the one hand, such systems can be unique allies in the protection of critical infrastructure. On the other hand, the same technology could in theory be used to develop more sophisticated attacks or automate cyber offensives, a scenario that recalls the dangers of a large-scale fraud facilitated by AI(/noticias/fraude-bilionaria-ia-ex-ceo-cfo-acusados-fabricar-clientes-receitas-ficticias).

Anthropic's decision to stop the release and promote a public dialogue signals an attempt to establish new parameters for the launch of high-impact AI technologies. The success of this strategy will depend on the active participation of regulators, ethics experts and the digital security community. Meanwhile, Mythos becomes a symbol of the dual potential of AI: a powerful tool to defend the digital world, whose control requires extreme responsibility and transparency.

Source: Digital Journal - Anthropic says will put AI risks ‘on the table’ with Mythos model