In an era where artificial intelligence is no longer a future promise but a daily operational reality, Jay Chaudhry, Chairman, CEO, and co-founder of Zscaler, appeared on Bloomberg's "The Close" to sound the alarm. As we navigate April 2026, the release of advanced models such as Anthropic's Mythos AI has fundamentally reshaped the threat landscape, rendering traditional defense methods not just inadequate, but dangerously obsolete.
The New Reality of AI-Driven Threats
According to Chaudhry, demand for Zscaler's services has never been higher, and the reason is simple: AI has "democratized" cybercrime. With tools like Mythos, even malicious actors with limited technical expertise can now create highly convincing phishing attacks, automated malware that mutates in real-time, and complex social engineering strategies. The speed at which these threats evolve far exceeds the ability of human security teams to react manually.
Chaudhry emphasized that Mythos AI, despite its built-in safeguards, can be used indirectly to analyze security gaps in corporate networks. "It's no longer about a hacker trying to crack a password," he noted. "It's about a machine testing thousands of doors simultaneously, finding the single weakness in seconds." This asymmetry between attacker and defender is what is driving large enterprises toward cloud-native security solutions.
The Collapse of the 'Moat and Castle'
For decades, cybersecurity relied on the castle model: a strong wall (firewall) protecting everything inside. Chaudhry explains that in the age of cloud and remote work, this model has completely collapsed. Employees now use hundreds of SaaS applications and AI tools that reside outside the traditional corporate network. Zscaler promotes a "Zero Trust" architecture, which does not rely on the user's location but on identity and connection context.
- Identity over Network: Access is granted at the application level rather than the network level, preventing the lateral movement of attackers.
- AI Inspection: Using AI to fight AI—detecting anomalies in user behavior in real-time.
- Data Protection: Preventing the leakage of sensitive information to public AI models like Mythos.
The Challenge of Anthropic and LLMs
The Bloomberg discussion also focused on how businesses are struggling to balance the productivity offered by Mythos AI with security requirements. Many companies fear that their employees will input proprietary code or sensitive financial data into these models for analysis, exposing the company to massive risks. Chaudhry argued that the solution is not to ban these tools—which would hurt competitiveness—but to create "secure conduits" through which data is filtered before reaching the AI.
"AI is the ultimate accelerator. If you don't have a security infrastructure that moves at the same speed, you're already exposed," Chaudhry stated.
Zscaler is investing heavily in integrating its own machine learning models that analyze over 300 billion transactions daily. This massive volume of data allows the platform to identify attack patterns before they even become widely known in the cybersecurity community.
Economic Outlook and the Future
Despite an uncertain macroeconomic environment, the cybersecurity sector appears to remain resilient. Information security budgets (CISO budgets) are the last to be cut, as the cost of a successful cyberattack—in reputation, fines, and lost productivity—is now existential. Chaudhry closed the interview by expressing optimism for Zscaler's growth, noting that the transition from legacy hardware to cloud security is still in its early stages.
The conclusion is clear: The rise of Mythos AI and its peers is not just a technological evolution, but a catalyst forcing every organization to rethink the foundations of its digital existence. "Zero Trust" is no longer an option, but a survival necessity in the 21st century.
