Small Firms Race to Defend Against Ai-powered Cyberattacks

WHO:

Inc - The article identifies small businesses as “low-hanging fruit” for AI hackers and points readers to Inc. Power Partner companies for cybersecurity solutions.

Fast Company - Bartley Richardson is CrowdStrike’s chief AI and autonomous systems officer; David Cooper is EY’s cyberstrategy activation leader; Mona Rahimi is an associate professor of software engineering in Northern Illinois University’s Department of Computer Science.

Fast Company - Joni Klippert is StackHawk’s cofounder and CEO; Scott Gerlach is its cofounder and chief security officer.

Fast Company - Christopher Robinson is Akrites’s CTO and chief architect and works for the Open Source Security Foundation; Ami Luttwak is Wiz’s cofounder and CTO.

Fast Company - OpenAI, Microsoft, and Anthropic have launched projects to probe and harden digital systems. CrowdStrike, EY, StackHawk, the Linux Foundation, Akrites, Wiz, and Google DeepMind are also named in the article.

WHAT:

Fast Company - The article reports that modern AI systems can find software vulnerabilities and work around cybersecurity defenses. Training exercises involving models from OpenAI and Anthropic resulted in unintended real-world breaches; the companies restricted access to their most capable models.

Fast Company - Cooper says clients’ security operations centers and telemetry show a “measurable increase in the volume and velocity” of attacks. The article says businesses are seeing more cyberattacks, at least partly attributed to AI assistance.

Fast Company - CrowdStrike announced SafeMind in September. It uses custom AI models to probe simulated customer infrastructure, while defensive AI detects attacks, counters them, and identifies security holes. The systems learn from repeated exercises; SafeMind records environmental changes and gives the log to the offensive harnesses.

Fast Company - Companies Cooper works with have essentially universally deployed AI to probe and patch systems. Microsoft reportedly unveiled its largest security update in history.

Fast Company - Cybersecurity Ventures estimated global cybercrime costs at $10.5 trillion in 2025. A World Economic Forum report warned that “AI is supercharging the cyber arms race.”

Fast Company - StackHawk tests running applications for vulnerabilities. Klippert calls the increase in fixes “an explosion in rapid remediation.”

Fast Company - Akrites, announced by the Linux Foundation and industry partners in June, coordinates open-source security reports and provides documentation, examples, and sometimes patch code. It can also release a fork and seek a “maintainer of last resort” when a project lacks an active maintainer or cannot accept patches.

Fast Company - Wiz’s Scan for Good initiative, in collaboration with Google DeepMind, uses AI to find and fix vulnerabilities for critical infrastructure organizations, governments, healthcare institutions, and nonprofits.

WHERE:

Fast Company - Akrites prioritizes open-source projects used in critical infrastructure, including financial services and telecommunications.

WHY:

Fast Company - AI can chain minor software flaws into pathways for unauthorized access and work persistently to find vulnerabilities. Defenders use AI to identify and repair flaws more quickly.

Fast Company - Security teams often require human oversight because AI can make mistakes and security changes can disrupt critical services. Rahimi says, “We cannot really maximize security at any cost,” because services must continue operating.

Fast Company - AI-generated reports and proposed fixes have burdened open-source maintainers, many of whom are unpaid volunteers. AI coding tools can also introduce bugs.

HOW:

Fast Company - AI hacking agents can probe live systems, refine attacks, and combine vulnerabilities into complex exploit sequences, including zero-day attacks.

Fast Company - Defenders can use proprietary source code and infrastructure details to help AI probe and repair systems. Humans may review significant changes or balance security with service availability.

Fast Company - Luttwak says, “This concept of static security controls is dead,” describing protection as a “cat-and-mouse game” requiring companies to conduct continuous AI scans.

(UPDATED October 21, 2026 01:31 UTC)

AI is accelerating a cybersecurity arms race: modern systems can uncover software vulnerabilities, chain minor flaws into attacks, and even work around defenses, while security teams are also using AI to find and fix weaknesses faster. CrowdStrike’s SafeMind uses custom AI to test simulated customer infrastructure, while projects from OpenAI, Microsoft, Anthropic, the Linux Foundation, and others aim to probe and strengthen digital systems; Wiz and Google DeepMind are also using AI to help protect critical organizations. But experts say human oversight remains essential because AI can make mistakes, proposed fixes can burden open-source maintainers, and security changes must not disrupt vital services.