AI Models Expose Critical Flaws in Software Used by Most Websites, NSA Warns
OpenAI and Google AI models have helped researchers find serious security problems in software used by most websites around the world. This has led the U.S. National Security Agency to warn about major cybersecurity risks.
OpenAI and Google AI models have helped researchers find serious security problems in software used by most websites around the world. This led the U.S. National Security Agency (NSA) to warn about major cybersecurity risks.
The findings show that artificial intelligence (AI) can help people find weaknesses in widely used software. However, the report does not name the affected products, explain the problems or say whether attackers have used them.
A major cybersecurity concern
Security problems in software can put websites at risk if attackers find and use them. How widely the software is used matters. A problem in a tool used across much of the internet could affect many organizations. However, the available information does not show how many websites may be at risk.
According to the report, researchers used AI models from OpenAI and Google to find these serious problems. The report does not say which models they used, how they tested the software or whether either company took part in the research beyond providing the models.
The NSA's warning shows that the issue is important for national cybersecurity. The agency has described the security problems as critical. However, the available information does not include its technical advice, suggested fixes or a deadline for software providers to take action.
How AI helps find security problems
AI systems can help researchers examine computer code and spot possible weaknesses. In this case, the reported discovery shows how important this work can be when AI tools are used to check software that supports a large part of the internet.
Finding a security problem does not prove that anyone has used it in an attack. The report gives no details about harmful activity, the researchers' identities or the software companies involved. It also does not say whether fixes are available.
This leaves website operators with important questions. They do not yet know which products need attention, how quickly they should act or what they should do to protect affected systems. The NSA's warning shows that the problems are serious, but more technical details are needed to understand the actual risks and decide what action is necessary.
Source: ft.com