AI firms position themselves as cybersecurity providers amid rising software vulnerability crisis

Recent high-profile security incidents, including the Binance hack and an attack averted at Anthropic, highlight growing digital infrastructure risks. These events are occurring as generative AI companies like OpenAI and Anthropic simultaneously empower hackers while positioning themselves as the primary defense. This trend suggests a shift toward AI firms operating as a form of protection racket. The threat landscape is substantial, with approximately 400,000 documented Common Vulnerabilities and Exposures (CVEs) since 1988, including 13,000 in the last month. A critical PHP language bug, PHP2Shell, demonstrated this danger, prompting some websites, like Wikimedia, to temporarily shut down. While the PHP Foundation was aware of the bug in 2024, the threat escalated significantly with the advent of generative AI. In response, AI developers are actively entering the cybersecurity market. OpenAI launched the Patch the Planet initiative and unveiled GPT-6 Cyber, while Anthropic offers Claude Mythos Preview, and Google has Gemini Carbon 4. Sam Altman does not believe all bugs can be patched, and Jason Clinton of Anthropic has termed the situation a vulnerability tsunami, signaling years of instability. These companies are monetizing this risk, with access to models like OpenAI's Daybreak program being subsidized, though long-term pricing increases are anticipated.