Artificial intelligence has stopped being only an automation or content generation tool to become a central actor within global cybersecurity. In 2026, one of the most relevant cases in this scenario is that of Anthropic, the artificial intelligence company behind Claude, which surprised the technology sector with the development of Mythos, a model specialized in detecting software vulnerabilities at large scale. The company recently announced that it will allow partners in its cybersecurity program to share with third parties the findings obtained through this technology, marking an important strategic change in the way digital threats are managed.
The decision has deep implications for governments, banks, technology companies, software developers, and security experts. It also opens a debate about the balance between defense and risk: a tool capable of finding thousands of critical flaws could help protect digital infrastructures, but it could also become an extremely powerful weapon if it falls into the wrong hands.
The Birth of Mythos and the Glasswing Project
Anthropic introduced Mythos as part of an initiative driven by Anthropic called Project Glasswing, a restricted Anthropic program focused on defensive cybersecurity. The system developed by Anthropic uses advanced AI models capable of analyzing enormous amounts of code and locating complex vulnerabilities that even specialized human teams had overlooked for years. With this project, Anthropic sought to position itself as one of the most influential companies in the field of artificial intelligence-based cybersecurity.
Unlike traditional AI models focused on productivity or text generation, Mythos, created by Anthropic, was specifically designed for offensive and defensive tasks within the field of cybersecurity. According to various reports, Anthropic’s model can identify programming errors, generate exploitation proofs, and detect weaknesses in operating systems, web browsers, and enterprise applications. Thanks to these capabilities, Anthropic began attracting the attention of technology companies and organizations concerned about the increase in sophisticated digital threats.
Anthropic’s Glasswing program initially brought together around forty selected organizations, including major technology companies and financial institutions. Amazon, Microsoft, Nvidia, and Apple participated in testing the system developed by Anthropic, while major U.S. banks began using tools linked to Anthropic to detect possible weak points in their digital infrastructures. Collaboration with Anthropic allowed these companies to explore new ways of automating security audits and strengthening their systems against possible attacks.
From the beginning, Anthropic maintained strong restrictions on the use and disclosure of discoveries made with Mythos. Although there was no absolute formal confidentiality agreement, Anthropic did establish important limitations to avoid leaks of sensitive information. The company feared that premature public exposure of certain vulnerabilities found by Anthropic could facilitate massive cyberattacks and create risks for critical infrastructures in different technological and financial sectors.
However, with the progress of the program and growing pressure from the technological ecosystem, Anthropic reconsidered its position. Finally, Anthropic decided to allow part of the findings related to Mythos to be shared with third parties under certain responsible disclosure criteria, a measure that reflected Anthropic’s change of strategy in the face of global cybersecurity challenges.
The Policy Change: Sharing to Protect
In May 2026, Anthropic announced that it would allow Glasswing participants to share findings, tools, and practices derived from the use of Mythos with other potentially affected organizations. The measure promoted by Anthropic represents a significant change compared to the initial strict control strategy that Anthropic had maintained since the beginning of the project. With this decision, Anthropic sought to promote greater collaboration among technological actors and organizations dedicated to cybersecurity.
The logic behind this decision made by Anthropic is simple: if a critical vulnerability affects multiple companies or sectors, keeping it secret may be more dangerous than coordinating a rapid and collective response. Anthropic argued that supporting the responsible exchange of information strengthens global defensive capacity against growing threats driven by artificial intelligence. In addition, Anthropic believes that sharing certain findings in a controlled manner can help prevent large-scale attacks and improve the preparedness of different industries.
The new approach developed by Anthropic will allow program partners to collaborate with regulators, government agencies, open-source software communities, incident response teams, and other vulnerable companies. Anthropic emphasized that these disclosures must follow “responsible disclosure” principles, that is, controlled notifications that allow flaws to be corrected before they are publicly exploited. According to Anthropic, this model seeks to balance transparency with the need to avoid risks derived from premature exposure of critical vulnerabilities.
This movement driven by Anthropic reflects a cultural change within the AI industry. Until a few years ago, many technology companies tended to keep their most advanced capabilities under strict secrecy, including developments similar to those of Anthropic. Today, the accelerated growth of digital threats forces the adoption of more collaborative models, and Anthropic is trying to position itself as one of the references of this new strategy of technological and defensive cooperation.
An AI Capable of Finding Invisible Vulnerabilities
What has made Mythos, developed by Anthropic, such a relevant topic is not only its technical sophistication, but also the results obtained by Anthropic during system testing. According to reports related to Anthropic’s project, the system discovered thousands of important vulnerabilities in widely used software. These discoveries increased international attention on Anthropic’s capabilities within the field of advanced cybersecurity.
Mozilla, for example, managed to fix more than 400 flaws in Firefox thanks to tools associated with Mythos and investigations driven by Anthropic, including errors that had remained hidden for more than fifteen years. Indirect collaboration with technologies developed by Anthropic made it possible to accelerate audit processes and improve the detection of complex threats in widely used browsers.
The findings related to Anthropic and Mythos were not limited to browsers. Anthropic stated that the AI identified critical weaknesses in practically all major operating systems and software platforms analyzed by Anthropic during the development of the program. This level of detection turned Anthropic into one of the most closely watched companies within the debate on artificial intelligence applied to cybersecurity.
The potential impact of Anthropic’s investigations is enormous because the global digital infrastructure depends on legacy systems that, in many cases, contain code developed decades ago. Numerous financial companies, hospitals, airports, public services, and government agencies continue using old technologies that are difficult to update, which increases the relevance of tools created by Anthropic to identify hidden flaws before they can be exploited.
The capability of an AI developed by Anthropic to examine millions of lines of code and detect anomalous patterns at speeds impossible for humans radically changes the rules of the game. Thanks to these capabilities, Anthropic has become one of the main protagonists of technological transformation within the global cybersecurity sector.
The Fear of a Digital Arms Race
Despite its defensive benefits, Mythos and Anthropic’s advances have also raised concerns among governments and security specialists. The main fear is that tools similar to those created by Anthropic could be used by criminal groups, hostile state actors, or ransomware networks to discover vulnerabilities before they can be corrected. This scenario has generated international debates about the risks associated with the rapid advancement of technologies driven by Anthropic and other artificial intelligence companies.
For years, the process of finding complex vulnerabilities required highly specialized knowledge and a great deal of manual analysis time. However, technologies developed by Anthropic threaten to drastically reduce that technical barrier, allowing the automation of tasks that previously could only be performed by highly trained experts. This technological change associated with Anthropic particularly concerns organizations dedicated to digital defense and the protection of critical infrastructures.
Some experts fear that the advances driven by Anthropic could provoke an “industrialization” of cyberattacks, where advanced models automate tasks that previously could only be performed by experienced hackers. In this context, the evolution of systems similar to Mythos could profoundly modify the balance between attackers and defenders within the global digital ecosystem.
Concern increased when it became known that Mythos, developed by Anthropic, could generate functional exploitation proofs and detect complex chains of vulnerabilities. Although Anthropic insists that it maintains strong access controls over these tools, the mere fact that a technology with such capabilities exists already generates concern among regulatory agencies, governments, and digital defense teams closely monitoring Anthropic’s advances.
Banks and Systemic Risk
One of the sectors most alarmed by the advancement of these tools developed by Anthropic has been the financial sector. Various U.S. banks began using Mythos, the system created by Anthropic, to review their internal systems after warnings emerged about possible critical vulnerabilities detected through technologies associated with Anthropic. Anthropic’s growing influence within the financial field reflects the extent to which banking institutions consider strengthening their digital defenses against increasingly sophisticated threats a priority.
The Financial Stability Board (FSB), the international organization responsible for supervising global financial stability, requested meetings with Anthropic to better understand the implications of the model developed by Anthropic. Conversations with Anthropic were aimed at evaluating the potential impact that tools like Mythos could have on international financial systems and on global economic stability in the event of coordinated attacks or massive exploitation of vulnerabilities.
The concern related to Anthropic is not exaggerated. The international financial system depends on extremely interconnected technological networks, many of them based on complex infrastructures that could be analyzed by technologies similar to those developed by Anthropic. A vulnerability exploited in a coordinated way could affect international payments, stock exchanges, banking compensation systems, or critical liquidity infrastructures, generating large-scale economic consequences.
The message emerging from Anthropic’s advances is clear: digital security is no longer just a technical problem, but an issue of global economic stability. The growing sophistication of tools such as Mythos has led governments, banks, and international organizations to consider Anthropic one of the most influential actors within the recent evolution of artificial intelligence-based cybersecurity.
The Pentagon and National Security
Interest in Mythos and Anthropic’s investigations also reached the military field. The United States Department of Defense began using the model developed by Anthropic to detect vulnerabilities within government systems and critical networks related to national security. Anthropic’s participation in this type of initiative demonstrates the extent to which artificial intelligence has become a strategic priority for military and defense organizations.
The Pentagon’s involvement in projects linked to Anthropic demonstrates how much artificial intelligence is transforming national cybersecurity. Digital threats are no longer considered only economic crimes or isolated attacks, but strategic risks capable of affecting defense, energy, telecommunications, and public security. In this context, Anthropic’s technological advances acquire relevance not only for private companies, but also for governments and national security agencies.
Within this scenario, the AI developed by Anthropic becomes a double-edged sword. On one hand, Anthropic’s technologies make it possible to strengthen defenses, automate massive audits, and improve the detection of critical vulnerabilities. On the other hand, specialists warn that tools similar to those of Anthropic could increase the potential sophistication of future cyber offensives if they were used for malicious purposes.
The challenge for governments and companies will be to find control mechanisms that allow them to take advantage of the defensive capabilities driven by Anthropic without facilitating a scenario of uncontrolled proliferation of advanced cybersecurity technologies. As Anthropic continues developing increasingly sophisticated tools, the debate over regulation, access, and international supervision will continue to grow.
The Rise of “Offensive AI”
For years, the technology industry mainly talked about generative AI applied to productivity, virtual assistants, or content creation. However, the case of Anthropic and Mythos shows the emergence of a new category: specialized offensive and defensive AI. Thanks to initiatives such as those of Anthropic, artificial intelligence began taking a much more active role within advanced cybersecurity.
The systems developed by Anthropic not only answer questions or write code. They can also analyze complex digital architectures, detect unsafe behaviors, and generate hypotheses about possible attack vectors. These capabilities make Anthropic one of the main references within a new generation of technologies specifically focused on offensive and defensive cybersecurity.
In practical terms, Anthropic’s advances mean that a process that previously could take weeks or months can now be completed in a matter of hours. Tools driven by Anthropic make it possible to automate audits and accelerate security investigations at speeds that were previously impossible for traditional human teams.
The consequences of Anthropic’s growth are profound for the cybersecurity industry. Audit firms, technology consultancies, and bug bounty platforms are already adapting to this new environment driven by artificial intelligence and technologies similar to those developed by Anthropic. Many organizations believe that Anthropic’s influence could redefine the way security testing and vulnerability analysis are carried out in the coming years.
Some specialists believe that human work will change instead of disappearing due to the advancement of systems developed by Anthropic. Researchers will continue to be necessary to interpret findings, validate risks, and understand the operational context of each vulnerability detected through tools associated with Anthropic. Human experience will continue to be key for making strategic decisions in the face of complex threats.
However, repetitive and massive exploration tasks could become widely automated thanks to technologies developed by Anthropic. This change driven by Anthropic could profoundly transform the cybersecurity labor market and modify the balance between automation and human analysis within the global technology industry.
The emergence of Mythos developed by Anthropic marks one of the most important moments in the recent evolution of cybersecurity. Anthropic not only developed a tool capable of detecting vulnerabilities at large scale, but also opened a global debate on how these capabilities driven by artificial intelligence should be managed. Thanks to Anthropic’s advances, governments, technology companies, and financial organizations began rethinking their digital protection strategies against increasingly sophisticated threats.
The decision made by Anthropic to allow partners to share findings related to Mythos represents recognition that digital defense requires collaboration and speed in the face of growing threats. Anthropic understood that keeping security discoveries completely isolated could limit the response capacity of other vulnerable organizations. For this reason, Anthropic began promoting a more collaborative approach that allows strengthening the technological resilience of different critical sectors.
However, the advances driven by Anthropic also expose a troubling dilemma: the same tools developed by Anthropic that can protect critical infrastructures could become extremely dangerous instruments if used for offensive purposes. This scenario generates concern among specialists, governments, and international organizations that observe how technologies similar to those created by Anthropic could alter the global cybersecurity balance.
The story of Mythos and Anthropic illustrates a profound change in the relationship between artificial intelligence and cybersecurity. It is no longer simply about automating tasks or generating text through AI; now, tools developed by Anthropic directly participate in the identification of flaws that could affect banks, governments, military systems, and essential services. Anthropic’s growth within the technology sector demonstrates how artificial intelligence is becoming a strategic component for the protection of critical digital infrastructures.
infrastructures. The technological world is thus entering a new stage where digital security will increasingly depend on intelligent systems such as those developed by Anthropic, capable of finding vulnerabilities before attackers do. The central question is no longer whether this transformation driven by Anthropic will occur, but who will be capable of controlling it and under what rules the future of artificial intelligence applied to cybersecurity will develop.
In this context of technological transformation and growing digital threats, having specialized allies in cybersecurity and technological infrastructure becomes increasingly important. ITD Consulting offers advanced solutions in cybersecurity, digital transformation, infrastructure management, and data protection for companies seeking to strengthen their operations against the challenges of the modern digital environment. To learn more about ITD Consulting’s services and receive specialized advice, you can write to [email protected].
EN 
ES