The AI Security Loop: Vendors Warn of Risks They Helped Create
AI-generated image · US National Wire
A coalition of over 100 tech giants, including OpenAI and Microsoft, warns that current cybersecurity is failing—and their proposed fix is more AI.
For critical infrastructure operators, the tech stack has always been a balancing act between innovation and vulnerability. Now, as *The Register* first reported, the industry is attempting to pivot that risk into a new product category.
According to reporting from *The Register*, more than 100 of the world's largest technology and information security firms have signed an open letter warning that the current approach to cybersecurity is no longer effective. The signatories include the very architects of the current AI boom—OpenAI, Google, Microsoft, and Anthropic—alongside cloud and enterprise giants like AWS, IBM, Oracle, and Cisco.
**Opinion:** From an infrastructure operator's perspective, this is a classic industry pivot. The same vendors who spent years selling the 'status quo' of cloud services and security software are now sounding the alarm that those very systems are insufficient. We are seeing a systemic risk being rebranded as a recurring SaaS opportunity.
*The Register* notes that the coalition, which also includes security heavyweights such as CrowdStrike, Palo Alto Networks, Fortinet, and Cloudflare, warns that AI-enabled attacks will become "far more widespread and sophisticated" in the coming months. The letter specifically highlights risks to internet infrastructure, water treatment plants, and hospitals.
While the letter acknowledges that long-standing issues—such as unpatched software, weak authentication, and misconfigurations—have plagued critical infrastructure for years, the proposed remedy is not a fundamental redesign of security. Instead, the solution is, as *The Register* reports, more AI.
The coalition's roadmap includes: * Deploying cheaper AI models to handle security tasks at scale. * Reserving "frontier systems" for more complex security problems. * Encouraging security vendors to continuously test defenses against frontier AI capabilities. * Calling on governments to fund cybersecurity for essential services.
Despite the urgency, *The Register* points out a glaring omission: the letter contains no firm deadlines, no specific dollar amounts for the "significant funding" requested from model developers, and no concrete commitments from the signing companies.
As AI agents begin finding and exploiting vulnerabilities independently and AI-generated exploit code appears in attacks, the industry is urging defenders to act now. However, as *The Register* highlights, the industry's ultimate fix for the AI security problem is simply more resources and more AI—leaving the question of who will actually foot the bill unanswered.