Supporting a global effort to put AI in the hands of defenders, strengthen software security and verify that fixes actually work
Pervaziv AI is proud to join OpenAI and organizations across technology, cybersecurity, government and industry in signing the Open Letter for a Global Surge in Cyber Defense.
The initiative comes at an important moment. OpenAI argues that rapidly advancing AI capabilities are creating a limited window in which defenders can use the same technology transforming the threat landscape to address longstanding vulnerabilities, insecure software, excessive permissions, misconfigurations and other weaknesses across digital infrastructure. Pervaziv AI is now listed among the initiative’s signatories.
For Pervaziv AI, that mission closely reflects the journey we have been pursuing with Cortex.
Building AI Around Defense
Cortex began as an AI coding and security agent inside Visual Studio Code, bringing vulnerability identification, security scanning and remediation closer to developers. Over the past year, it has evolved into a broader Enterprise AI Control Layer spanning specialized models, AI agents, security and privacy controls, multicloud environments, enterprise systems, browsers, IDEs and mobile experiences.
Security has remained part of that architecture rather than becoming a separate layer added later.
Cortex 4.2 introduced AI Threat Model and AI Security Review, bringing architectural threat analysis and implementation-level security reasoning directly into development workflows. Cortex 5.2 expanded that approach through a specialized AI Model Ensemble combining on-device privacy and prompt-injection defenses with security analysis, safety-aware reasoning and coding intelligence.
More recently, Cortex Verify added an independent validation layer designed to evaluate whether proposed software changes satisfy requirements, follow standards and have sufficient testing and evidence before moving forward.
That distinction becomes increasingly important as AI writes more software.
From Finding Problems to Verifying Outcomes
OpenAI’s collective cyber defense initiative calls on organizations to fix their highest-risk weaknesses, raise security standards for what they build and deploy, including AI-generated code, and verify that remediation actually works. It also calls on cybersecurity companies and technology partners to make AI-powered defense more accessible and continuously strengthen defenses against increasingly capable attacks.
Those principles align closely with how we think about the next generation of secure AI-assisted engineering.
Finding a vulnerability is useful. Generating a patch is useful. But neither is enough on its own.
Defensive AI ultimately needs to connect discovery, reasoning, remediation, testing and verification while preserving the security boundaries and human accountability surrounding consequential decisions.
“As AI becomes more capable on both sides of cybersecurity, defense cannot remain fragmented or reactive. We need to help more defenders find weaknesses earlier, remediate them faster and verify that the fixes actually work. That is why collective cyber defense matters, and why Pervaziv AI is proud to support this effort.”
Anoop Jaishankar, Founder and CEO, Pervaziv AI
A Collective Opportunity
Recent Cortex work around specialized routing and long-running context reflects the same broader direction. Different requests can be coordinated across models, search and engineering skills, while semantic context compaction helps preserve important facts, evidence and validation across increasingly complex AI workflows.
The technology will continue to evolve. So will the threats.
What matters is ensuring that advances in AI translate into a stronger advantage for defenders.
Pervaziv AI is pleased to stand alongside OpenAI and the wider community supporting this initiative, and to continue building toward a future where AI helps organizations identify risk earlier, build more securely, verify outcomes and strengthen cyber defense at scale.
Read more on OpenAI’s collective cyber defense page.
