AI Models Expose Critical Flaws in Software Used by Most Websites, NSA Warns
OpenAI and Google AI models have helped researchers uncover severe security flaws in software used by most websites worldwide, prompting a warning from the U.S.
OpenAI and Google AI models have helped researchers uncover severe security flaws in software used by most websites worldwide, prompting a warning from the U.S. National Security Agency about critical cyber vulnerabilities.
The findings show how artificial intelligence can help identify weaknesses in widely deployed software. But available reports do not name the affected products, describe the flaws or say whether attackers have exploited them.
A broad cybersecurity concern
Software vulnerabilities can leave websites exposed when attackers find and exploit them. The scale of use matters: a fault in a tool deployed across much of the web could affect many organizations, though the available details do not establish how many sites are at risk.
Researchers used models from OpenAI and Google to detect the severe faults, according to the source report. It does not specify which models were involved, how the researchers tested the software or whether the companies took part in the research beyond providing the models.
The NSA’s warning puts the issue in a national cybersecurity context. The agency has identified the vulnerabilities as critical, but available reports do not include its technical guidance, any recommended fixes or a timetable for software providers to act.
AI’s role in finding flaws
AI systems can assist researchers in examining code and spotting potential weaknesses. In this case, the reported discovery underscores the stakes when such tools are applied to software that underpins a large share of websites.
Finding a vulnerability is not the same as proving that it has been exploited. Available reports offer no details about malicious activity, the identities of the researchers or the software vendors involved. It also does not say whether patches are available.
That leaves key questions unanswered for website operators: which products need attention, how urgently they should respond and what steps would secure affected systems. The NSA’s warning signals the seriousness of the flaws; further technical details will be needed to determine the practical risk and the response required.
