Nobel Prize-winning economist Paul Krugman warns that progress in artificial intelligence could make the internet as we know it nearly unusable within a very short time frame of just a few months, rather than in a distant future. Krugman highlights the risk that the AI boom could become a speculative bubble, whose collapse might push the U.S. economy into a new liquidity trap, while massive investments in data centers consume resources that could have been allocated to other economic sectors, generating relatively few jobs. Alongside these economic concerns, Krugman mentions warnings from the company Anthropic regarding the potential of AI to lead to the extinction of humanity; while alarming, this scenario seems somewhat counterbalanced by financial interests in certain contexts, such as the company's initial public offering (IPO). The central concern is the degradation of internet quality: critical vulnerabilities in older programming languages like PHP have been exposed because developers publicly release patches, allowing AI-driven hacking teams to quickly identify flaws and exploit systems before most users can update. This phenomenon has already begun: on Wednesday, October 28, multiple government websites in Europe and other regions became inaccessible following the discovery of vulnerabilities in old codebases, forcing administrators to take them offline temporarily. Costin Raiu (co-founder of The Linux Foundation Project TLP) describes the situation as a race to exploit vulnerabilities, predicting widespread compromise of vulnerable servers next, with security experts forecasting several years of periodic instability. Analyst Azeem Azhar refers to these events as the 'Wuhan moment for the internet,' comparing the severity of the situation to the impact of the COVID-19 pandemic on the global economy, while Krugman explains the mechanism: the complexity of old codes (PHP from the 1990s) and the inability of human hackers to analyze them efficiently makes AI a superior tool for breaking security. Azhar emphasizes that safety measures embedded in AI models, known as 'ablation,' can be easily bypassed by modifying key instructions. As an alternative solution to the fully open internet, 'walled gardens'—partially closed systems with strict verification—are proposed. The economist concludes that the concrete risk in the coming years is not unrealistic economic growth, but rather that AI could make us poorer by providing attackers with powerful tools and forcing society to adapt rapidly to threats for which it is unprepared. The economic impact could be significant: transitioning to a secured digital system involves forfeiting a portion of GDP growth (estimated at 10% attributed to the open internet), but the major issue remains the short-term transition, where sectors such as tourism or financial services would need to adapt radically. The process of declining internet quality and increasing vulnerabilities is already underway, necessitating economic and security reconfigurations to avoid the negative impact of escalating automated cyberattacks. [1]
- Categories:
- Technology
- Entities:
