El error más grande en ciberseguridad: algo que parece práctico, pero pone en riesgo toda su información personal
The article reveals that using AI to generate passwords creates a dangerous illusion of security. While large language models produce keys that appear complex by following standard formatting rules, they ultimately rely on learned patterns rather than true randomness. This predictability makes the generated passwords vulnerable to sophisticated cracking techniques, as attackers can anticipate or replicate the models' output logic, effectively neutralizing the perceived strength of the credentials. Empirical evidence confirms that the majority of passwords generated by popular AI tools are easily compromised. Research indicates that an overwhelming proportion of these keys can be decrypted quickly using modern hardware, highlighting a significant failure in current AI-generated security practices. The core issue lies in the deterministic nature of these models, which fail to provide the unpredictable entropy required for robust digital defense, leaving users exposed to credential stuffing attacks when these weak keys are reused across multiple accounts. This finding is crucial for open data because it challenges the assumption that AI-driven solutions are inherently superior or neutral regarding security. It underscores the necessity for transparency in how AI systems are trained and deployed for sensitive tasks, advocating for open scrutiny of their limitations. Understanding these vulnerabilities is essential for developing trustworthy open-source security tools and policies, ensuring that community-driven developments do not inadvertently introduce systemic weaknesses that compromise user privacy and data integrity.
Source: semana.comPublished on 2025-05-10
Related news
- Fabrizio Peralta: “Se sabe que en la Bahía se venden datos personales”
- Synthetic data fill AI training gaps, but can’t replace real-world nuance, experts say
- From synthetic data to AI agents, new SAS Viya innovations boost speed, productivity and trust
- Investigation finds Alberta government non-compliant with freedom of information laws
- Two-thirds of Premium Bond holders never win a prize despite increase in prizes worth £50 and £100 - IFA Magazine