CERT-In Warning: Are AI Apps Really Dangerous?

The use of artificial intelligence (AI) technology is growing rapidly, but is every AI app safe? India’s national cybersecurity agency, CERT-In, has issued an advisory alerting people that not all AI apps are reliable and may have several technical flaws. The advisory warns that there are several security flaws in the design, training, and interaction of AI, which cybercriminals can take advantage of.

Use secondary accounts. CERT-In Warning on AI apps and the dangers associated with it.

CERT-In Warning on AI:

According to the agency, cyber attackers can take advantage of the popularity of AI apps to create fake apps. If someone inadvertently downloads them, malware may be installed on their phone, increasing the risk of personal data being stolen. To minimize this risk, CERT-In has suggested that while logging into AI apps. Also users should use anonymous accounts without revealing their real identity.

Risk of data tampering

AI models are vulnerable to several threats. The biggest threat among these is data poisoning, in which AI is confused by giving it wrong data, due to which it gives wrong or biased results. In an adversarial attack, cyber criminals can affect the AI model in such a way that it becomes incapable of making the right decision. Sensitive information can be extracted from the training data of AI through a model inversion attack. Using prompt injection technology, hackers can break the security wall of the AI model and force it to answer on their terms.

ALSO READ: Economy: Who Are FIIs? What Brings Them Back To India?

Agency’s advice to users CERT-In has warned that relying too much on AI apps can be dangerous, especially when it is related to legal or medical decisions. AI models also have the problem of hallucinations, which means that they can sometimes present wrong or fabricated information. If there is any deficiency in the data of the AI model, it can give wrong and misleading results. AI apps should be used only for general content creation and question-answering. But they cannot be considered reliable options for important decisions like judicial or medical ones.

ALSO READ: Stock Market Wrap: Key Gainers and Losers on BSE, NSE