Anthropic, the AI company behind the Claude chatbot, has developed Mythos, an artificial intelligence system capable of identifying thousands of software vulnerabilities in nearly every major operating system and web browser. Due to fears that malicious actors could exploit this capability, Anthropic has opted not to publicly release Mythos and is instead providing access exclusively to select large organizations.
Controlled access to mitigate cyber threats
Anthropic is collaborating with top technology and financial firms including Amazon, Apple, Cisco, JPMorgan Chase, and Nvidia through an initiative called Project Glasswing. The goal is to enable these companies to test Mythos internally and strengthen their cybersecurity defenses before the technology becomes publicly accessible or falls into the hands of hackers.
The company cautioned on its website that misuse of Mythos could have severe consequences for national security, public safety, and economic stability.
Concerns from security experts and government officials
Cybersecurity experts warn that AI’s ability to automate and accelerate hacking poses a significant threat. Alissa Valentina Knight, CEO of cybersecurity AI firm Assail, described the situation as an ongoing crisis requiring urgent preparation, noting that AI-powered attacks are far faster and more sophisticated than previous human-only cyberattacks.
Mythos’ threat has also drawn attention from U.S. federal officials. Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell recently met with major bank CEOs to discuss the risks related to AI-enhanced cyberattacks. Anthropic has briefed senior government and industry representatives on Mythos’s capabilities.
International Monetary Fund Managing Director Kristalina Georgieva has similarly expressed concern over growing cyber risks to the global financial system, underscoring the need for better protections and regulatory guardrails in the AI era.
AI’s dual role in cybersecurity
While AI systems like Mythos excel at detecting software bugs quickly and at scale—outperforming human capabilities—this strength doubles as a potential tool for hackers to find and exploit security flaws faster. Hackers already employ AI to conduct highly targeted phishing attacks and automate malware distribution, increasing the volume and precision of cyber threats.
Zach Lewis, chief information officer at a health sciences university, noted that AI enables attackers to create more convincing phishing campaigns that are harder to detect. The release of Mythos could accelerate the exposure of vulnerabilities, leading to a rise in cyberattacks until defenses catch up with rapid patching.
Industry speculation and Anthropic’s reputation
Some security observers speculate that Anthropic’s careful rollout of Mythos may serve marketing or business objectives ahead of the company’s planned initial public offering later this year. However, Anthropic has publicly emphasized AI safety and responsibility, and Project Glasswing aligns with its stated commitment to develop AI technology thoughtfully and securely.
Why it matters
The development of Mythos highlights the escalating cybersecurity risks posed by AI capable of identifying systemic vulnerabilities at unprecedented speed. As critical infrastructure and financial systems become increasingly digital and interconnected, limiting AI tools’ public availability to prevent exploitation represents a new front in national and economic security. The collaboration between Anthropic and major companies illustrates a proactive approach to managing AI-driven risks before they can be weaponized by cybercriminals.
Background
The cybersecurity landscape has been rapidly evolving with the introduction of AI-powered hacking tools. In 2025, the speed at which new AI capabilities were adopted by threat actors increased sharply, accelerating in 2026. Prior to Mythos, AI utilities were already used to enhance phishing, ransomware, and identity theft attacks, enabling both low-skilled and advanced hackers to conduct more efficient and high-volume operations. Mythos represents a significant leap in AI’s role in identifying underlying software weaknesses on a global scale.
Read more Business stories on Goka World News.