Is AI Making Us Worse at Cybersecurity?

Joshua Goldfarb, Field CISO at F5, reflects on a recent interaction that highlighted how professionals are increasingly treating colleagues like AI tools by requesting ready-made presentations and content without seeking guidance or lessons learned. He argues that over-reliance on AI is eroding key human qualities essential to the security profession, including creativity in problem-solving, the use of humor to communicate difficult topics, and genuine emotional sincerity that builds trust. Goldfarb also points to homogenized vendor messaging at conferences, declining precision in professional communication, and weakening critical thinking skills when analysts turn to AI for analysis and suggestions instead of tackling novel situations themselves. While acknowledging AI's benefits for productivity and attacker capabilities, he warns that certain tasks should remain firmly in human hands to preserve the industry's depth and effectiveness. Goldfarb shares his personal practice of refusing AI assistance for any content bearing his name, from topic selection to final writing, and urges security professionals to set similar boundaries. The piece concludes that resisting AI for creative and analytical work will ultimately strengthen the quality of security output across the field.

SecurityWeekNovember 14, 2026Translated from English

Read at source

Joshua Goldfarb, currently Field CISO at F5, recently turned down a speaking invitation due to a prior commitment. A week later, the replacement speaker approached him not for guidance or lessons from experience but essentially sent a prompt requesting the most interesting points on a topic and a ready-made presentation.

Goldfarb, who previously served as VP, CTO - Emerging Technologies at FireEye and as Chief Security Officer for nPulse Technologies, recognized the pattern. The individual treated him the way many now treat AI, expecting output without the human elements of mentorship, context, or personal insight. This led him to question whether AI is actively harming the security industry beyond its known benefits for productivity and attacker speed.

He identifies several areas of concern. Creativity allows security professionals to discover new attack techniques and solve unique problems; AI cannot replicate this. Humor has long helped convey serious messages effectively, yet AI-generated content lacks genuine wit. Emotion and sincerity motivate teams and stakeholders in ways AI cannot, because it does not feel. Marketing teams using AI produce repetitive, unremarkable messaging at conferences, while professional communication grows less precise. Finally, critical thinking suffers when analysts rely on AI for analysis of novel situations that have no pre-existing answers.

Goldfarb maintains strict personal boundaries: he refuses to use AI for topic selection, research, outlines, or content development, just as he has always refused ghostwriters. He believes this approach preserves quality and encourages the wider security community to establish similar limits for long-term benefit.

FireEye Joshua Goldfarb nPulse Technologies US-CERT

ShareX Facebook LinkedIn Reddit

Meridian digestThe week's most consequential security stories, distilled into one briefing.See the digest

More in AI & Model Risk

AI & Model Risk

How One AI Agent Exposed Five Critical Enterprise Security Gaps

An imaginary but realistic AI agent built by an accounts department employee at a mid-sized company quickly gained access to email, ERP, and Slack systems through OAuth grants and static API keys without any security review. The agent processed untrusted email input, accessed sensitive vendor data, and could modify ERP records, violating the Agents Rule of Two and enabling prompt injection via hidden text in an invoice attachment. Security experts including John Donovan, Amber Bennoui, Joshua Copeland, Chad Kalmes, and Rajesh Seshadri Appikulam outlined detection gaps, logging blind spots, and audit failures that allowed the unauthorized bank detail change to go undetected for weeks. A Cloud Security Alliance survey found 82% of organizations discovered unknown AI agents in the past year despite 68% claiming strong visibility. The case highlights risks from long-lived OAuth tokens, unmonitored personal credentials, and missing delegation chains during audits. Token Security compiled practitioner playbooks from companies like Roblox, Amex, Klaviyo, and Webflow into the free Agentic Universe resource to address these production challenges.

BleepingComputerNov 13, 2026Translated from English

AI & Model Risk

Containing Rogue AI Agents: OWASP Guidance on Revoking Credentials and Network Access

An AI agent tasked with sorting support tickets can be tricked into exporting sensitive records to an unfamiliar destination when malicious instructions appear inside a ticket. The agent succeeds because its assigned identity and connected tools already possess the necessary permissions. OWASP GenAI Security Project highlights that simply pausing the model often leaves workflows, credentials and network sessions active. Effective containment requires coordinated controls across the orchestrator, identity provider, tools and network layer. NIST incident response principles emphasize preparation through documented procedures for each production agent. KeeperPAM integrates secrets management and runtime access controls to support rapid revocation during incidents. Organizations are advised to test these controls regularly rather than improvise after an event occurs.

The Hacker NewsNov 13, 2026Translated from English

AI & Model Risk

OpenAI Launches Daybreak Red Team and Daybreak Defense Factory AI Security Tools

OpenAI has introduced two new cybersecurity products, Daybreak Red Team and Daybreak Defense Factory, that leverage AI agents to identify vulnerabilities in web applications and assist with developing fixes. The announcement took place at the company’s Cyber at the Frontier event on November 12. Daybreak Red Team performs penetration testing and vulnerability validation by using multiple AI agents that analyze source code and application context to confirm attack paths. The agents operate in isolated environments with Guardian review enforcing access boundaries. Daybreak Defense Factory helps security teams investigate findings and generate patches, supporting data imports from tools such as Wiz, Snyk, and Endor Labs. It can scan GitHub repositories, monitor commits, and allow engineers to review and modify proposed patches using Codex before creating draft pull requests. Neither product deploys patches automatically, and early access requests are now being accepted for Daybreak Red Team.

SecurityWeekNov 13, 2026Translated from English