SEC mandates and the AI attack surge
SEC rules require US-listed companies to disclose material cybersecurity incidents to the market on Form 8-K within four business days of the materiality determination. This disclosure must describe the nature, scope, timing, and impact of the incident. A delay in this reporting is only possible if the Attorney General finds that disclosure poses a substantial risk to national security or public safety. Foreign private issuers also face these requirements through Forms 6-K and 20-F, and they must use Inline XBRL tagging for these disclosures. The SEC rules for annual reports, specifically Regulation S-K Item 106, require companies to describe their risk management processes, management’s role, and the board’s oversight for all fiscal years that conclude on or after December 15, 2023. The Form 8-K Item 1.05 requirement also applies to smaller reporting companies as of June 15, 2024. These requirements apply to all companies subject to the reporting requirements of the Securities Exchange Act of 1934. I find the current regulatory pressure intense because a Palo Alto Networks and Google Cloud release found that 99% of respondents experienced at least one attack on their AI infrastructure over the prior year.
SASE convergence and Zscaler’s growth
Zscaler reports a 36x growth in AI/ML transactions in its security cloud, which equals a 3,000% increase in AI/ML traffic year-over-year. This massive increase in AI traffic changes where data lives and how attackers automate their work. Gartner projects that 60% of new SD-WAN purchases will be part of a single-vendor SASE offering by 2026, compared to just 15% in 2022. The convergence of technologies means that 60% of new SD-WAN purchases will be part of a single-vendor SASE offering by 2026, which is a massive increase from the 15% seen in 2022. This shift moves the industry away from point products toward platform decisions where SSE acts as the control plane and SD-WAN acts as the distribution layer. I see the market gravitating toward these platforms because the browser is the new branch for policy enforcement on unmanaged or BYOD endpoints.
| Metric | Value |
|---|---|
| Zscaler AI/ML transaction growth | 3,000% |
| SD-WAN single-vendor SASE projection (2026) | 60% |
| SD-WAN single-vendor SASE (2022) | 15% |
| AI infrastructure attack rate (Palo Alto/Google) | 99% |
| SEC 8-K incident disclosure window | 4 business days |
The SASE platform conversation is now table stakes for major vendors. The explosion of enterprise AI traffic, including SaaS AI apps, genAI platforms, and on-prem AI infrastructure, widens the blast radius. I find that the massive growth in AI/ML transactions makes the old conversation about network refreshes obsolete.
The death of rules-based DLP
Data security is the headline for modern SASE because DLP is moving from simple rules to context, identity, and posture across apps, web, and AI tools. Organizations add AI security / AI firewall controls to discover AI usage and enforce guardrails at runtime. I believe Zscaler’s AI-driven DLP matters now because legacy DLP fails against agentic workflows. These custom agents and SaaS AI apps operate completely outside traditional security visibility. The massive growth in AI/ML transactions proves that old-fashioned, rules-based data loss prevention is dead. I find the reliance on traditional rules a massive mistake when companies deal with agentic workflows that operate with high autonomy. You should know that the SASE operating model must control identity, data, and AI-driven workflows at cloud speed to prevent data leaks. I would skip any SASE implementation that ignores identity-based context. How can a company enforce guardrails if they cannot see the AI infrastructure?
