AI App Spending Report: Where Are the Security Tools?

TechTarget and Informa Tech’s Digital Business Combine.TechTarget and InformaTogether, we power an unparalleled network of 220+ online properties covering 10,000+ granular topics, serving an audience of 50+ million professionals with original, objective content from trusted sources. We help you gain critical insights and make more informed decisions across your business priorities.News, news analysis, and commentary on the latest trends in cybersecurity technology.An analysis of startup firms' spending on AI applications finds the top categories to be productivity and content-generation. Security? Not so much.November 4, 2025A recently released list of the top 50 artificial intelligence (AI) applications and services startups are actually paying for shows that they are, not surprisingly, focused on automation, productivity, and enterprise AI applications, with foundational model firms OpenAI and Anthropic leading the list.What's missing from the list? Security tools.The list, published in early October by venture capital firm Andreessen Horowitz (a16z) and based on data from fintech platform Mercury, shows that startups are investing in specific categories of capabilities, such as AI-powered development platforms — Replit (#3) and Cursor (#6) — and AI content-generation services — Freepik (#4) and ElevenLabs (#5). Mercury collected data from the aggregated expenditures of all its 200,000 customers.The data highlights a problem that has always been there but has simply shifted to a new crop of applications, says Melissa Ruzzi, director of AI at AppOmni, an enterprise software-as-a-service (SaaS) application security firm. Security is still not a top priority for these companies."What we're seeing here is really the 'build first and then worry about security later' [mentality]," Ruzzi says. "We see that even in terms of not just security but scalability [and] maintainability."Related:AI Security Agents Get Persona MakeoversWhile startups are their own microcosm in the business world — with extreme pressures to release products quickly — other companies can look at their lean operations and learn what to do and what not to do. Even though the list doesn't explicitly name security applications, that doesn't mean startups are disregarding security in their purchase decisions, says Zane Lackey, a general partner at a16z. "What we're seeing isn't that startups are ignoring security," he says. "It's more that security is increasingly embedded within the tools they use, from code generation to data management."Founders are moving faster than ever, trying to establish an AI-enabled business before competitors; part of that involves taking digital trust, data integrity, and platform security into consideration, Lackey says. AI-native security startups are already emerging that focus on development pipelines based on foundational models, verify the authenticity and provenance of training data, and detect malicious attacks on services, he says."It's early days, but we expect to see a new class of companies that are purpose-built for AI security move from the infrastructure tier into mainstream adoption as the ecosystem matures," Lackey adds.Related:Closing the AI Execution Gap in Cybersecurity — A CISO FrameworkYet the AI-usage story for startups mirrors that for cloud apps a decade ago. While startups expected cloud application providers to incorporate security into their products, SaaS providers' main goal continued to be features and functionality, not security, argues AppOmni's Ruzzi. As a result of these decisions, cloud breaches and service disruptions have become frequent problems."When they're using those Top 50 [apps], people expect that the security is part of the service, so they're not thinking about separate security tools for those applications," she says. "But that's not necessarily true."AI transcription tools and notetakers, for example, have taken off — both Otter.ai and Happyscribe grace the Top 50 list — but their use leads to an increasing amount of sensitive business conversations being stored online, often without appropriate security. Cybersecurity researchers have found numerous ways to bypass OpenAI's guardrails, while Replit deleted the production database of a firm using the platform for development. And because developers are producing more code with AI, the amount of security debt in those applications is growing, with secrets sprawl a major concern.At the very least, startup companies that offer a service or application need to consider their custodial responsibilities for the data they handle on behalf of their customers, says Ruzzi. Depending on the user agreement and government regulations, companies may not be able to put their customers' personal information into another AI service.Related:Let's Get Physical: A New Convergence for Electrical Grid SecurityDetermining security strategy means that startup workers should communicate with each other, she says. Different functional groups need to share information about which AI applications they plan to adopt and evaluate the security of those applications."The marketing person can ask the [IT/security] guy, 'Hey, what are the things I should be concerned about?'" Ruzzi says. "I think the key point for the startups is to use their own internal AI expertise not just to implement and develop AI, but also in the AI that they are using for the other departments."While the data shows that AI has become a significant component of how startups build and operate, they need to maintain strong security fundamentals, a16z's Lackey says."Founders should approach AI services with the same rigor they apply to any new technology they leverage," he says. "They still need to understand data flows, enforce access controls, and ensure model inputs and outputs are appropriately safeguarded."Robert Lemos, Contributing WriterVeteran technology journalist of more than 20 years. Former research engineer. Written for more than two dozen publications, including CNET News.com, Dark Reading, MIT's Technology Review, Popular Science, and Wired News. Five awards for journalism, including Best Deadline Journalism (Online) in 2003 for coverage of the Blaster worm. Crunches numbers on various trends using Python and R. Recent reports include analyses of the shortage in cybersecurity workers and annual vulnerability trends.2025 DigiCert DDoS Biannual ReportDigiCert RADAR - Risk Analysis, Detection & Attack ReconnaissanceThe Total Economic Impact of DigiCert ONEIDC MarketScape: Worldwide Exposure Management 2025 Vendor AssessmentThe Forrester Wave™: Unified Vulnerability Management Solutions, Q3 2025How AI & Autonomous Patching Eliminate Exposure RisksThe Cloud is No Longer Enough: Securing the Modern Digital PerimeterSecuring the Hybrid Workforce: Challenges and SolutionsCybersecurity Outlook 2026Threat Hunting Tools & Techniques for Staying Ahead of Cyber AdversariesYou May Also LikeAI Security Agents Get Persona MakeoversSora 2 Makes Videos So Believable, Reality Checks Are RequiredOperational Technology Security Poses Inherent Risks for ManufacturersAn 18-Year-Old Codebase Left Smart Buildings Wide OpenCopyright © 2025 TechTarget, Inc. d/b/a Informa TechTarget. This website is owned and operated by Informa TechTarget, part of a global network that informs, influences and connects the world’s technology buyers and sellers. All copyright resides with them. Informa PLC’s registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. TechTarget, Inc.’s registered office is 275 Grove St. Newton, MA 02466.