The blog post critiques the argument that prompt injection strings should be treated as zero-day vulnerabilities and not shared publicly. The author argues that these injections, although unpatchable, can be understood and mitigated through shared knowledge. The post emphasizes the value of transparency in AI security, suggesting that the concerns of making information public are overstated. The author critiques the notion of security theater, highlighting the speed of AI adoption and the need for robust defenses against evolving AI threats.