# The Defender's Window: Cyber security keynote **URL:** \ **Category:** Content Creators\ **Tags:** agents, open-ai, security, automation, openai\ **Created:** [28 September 2026 13:29 UTC](https://www.artofsm.art/t/the-defenders-window-cyber-security-keynote/24507 "2026-09-28T13:29:06Z")\ **Posts on this page:** 2\ **Page:** 1
**Author:** ![artesia](https://www.artofsm.art/user_avatar/www.artofsm.art/artesia/32/36_2.png) [@artesia](https://www.artofsm.art/u/artesia)\ **Post date:** [28 September 2026 13:29 UTC](https://www.artofsm.art/t/the-defenders-window-cyber-security-keynote/24507/1 "2026-09-28T13:29:06Z")
[![](https://www.artofsm.art/uploads/default/original/3X/f/d/fd2607af2e9f7aa1ad5aadbf649aa73414228743.jpeg "The Defender's Window: Cyber security keynote") ](https://www.youtube.com/watch?v=3jDhHA9JGUE) The keynote “The Defender’s Window: Cyber Security” highlighted OpenAI’s advancements in AI-driven cybersecurity, particularly through GPT-6 Astra and the “defense factory” initiative, which leverage AI agents to detect, validate, and remediate vulnerabilities efficiently while emphasizing safety and collaboration with governments and industry partners. Speakers showcased practical applications like automated vulnerability management, enhanced red teaming, and scalable security workflows, urging organizations to adopt AI tools promptly to maintain the current defensive advantage against cyber threats. ---
**Author:** ![artesia](https://www.artofsm.art/user_avatar/www.artofsm.art/artesia/32/36_2.png) [@artesia](https://www.artofsm.art/u/artesia)\ **Post date:** [28 September 2026 13:53 UTC](https://www.artofsm.art/t/the-defenders-window-cyber-security-keynote/24507/3 "2026-09-28T13:53:37Z")
The keynote titled “The Defender’s Window: Cyber Security” opened with Emanuel Mariel, General Manager for Europe, East, and Africa at OpenAI, emphasizing the rapid advancements in AI capabilities, particularly with the release of GPT-6 Astra. He highlighted the growing adoption of AI across industries, noting significant efficiency gains in businesses and over a billion weekly active users. However, with these advancements comes the critical issue of trust and safety, which OpenAI prioritizes heavily. Emanuel stressed the importance of a collective ecosystem approach involving company leaders, technology partners, and governments to address emerging cybersecurity threats, especially as AI models become more powerful and accessible. Matt, Head of Engineering for Cyber at OpenAI, introduced the concept of the “defense factory,” an internal initiative to leverage AI models for continuous cybersecurity defense. He explained how OpenAI mobilized cross-functional teams to strengthen their own defenses by using AI agents to identify and validate vulnerabilities dynamically. Matt underscored the urgency of acting during the “defender’s window,” a period where defenders currently have an advantage over attackers due to advanced AI tools. He also highlighted government collaborations, such as with Ukraine and the UK, to enhance national cybersecurity using these AI-driven approaches. Lena, leading OpenAI’s cyber program in Europe, detailed the capabilities of GPT-6 Astra, the most intelligent and aligned cyber security model to date. She explained how Astra improves vulnerability detection, including zero-day exploits, and enhances red teaming efforts. Lena showcased practical applications like the “Patch the Planet” initiative, which uses AI to secure widely used open-source software by assisting researchers in identifying and patching vulnerabilities rapidly. She also discussed the robust safety measures integrated into Astra to prevent misuse, including advanced monitoring and restrictions tailored to high-risk accounts. Vanessa, a cyber deployment engineer, demonstrated how OpenAI’s Codex Security tools integrate AI into practical cybersecurity workflows. She walked through scanning codebases for vulnerabilities, validating findings, and automatically generating patches using AI agents. Vanessa emphasized the importance of context in scans, such as enterprise-specific security concerns, and showed how Codex Security can streamline vulnerability management by automating ticket creation, patch submission, and team notifications. She also highlighted the scalability of these tools through command-line interfaces and SDKs, enabling organizations to manage security across thousands of repositories efficiently. Lou, OpenAI’s Field CTO for Cyber, shared insights into the internal implementation of the defense factory. He described how OpenAI uses isolated virtual environments to run AI agents that continuously scan, validate, assign ownership, and remediate vulnerabilities across their infrastructure. Lou presented impressive metrics, including a false positive rate below 1% and a 90% success rate in ownership assignment, demonstrating the effectiveness of this approach. He concluded by emphasizing OpenAI’s commitment to productizing these capabilities for customers, encouraging organizations to adopt AI-driven cybersecurity tools now to capitalize on the defender’s window and collaboratively enhance global digital security.