Government agencies should prevent the use of data for retraining when using artificial intelligence (AI) technologies and restrict AI's direct access to production databases and information systems of special importance, Elkhan Sadigli, Head of the Red Team and Artificial Intelligence Security Group at the State Service for Special Communication and Information Security, said at the 8th Summit of IT Heads of Government Agencies held in Baku, APA-Economics reports.
According to him, in addition to the agency's own requirements regarding artificial intelligence, international and European standards (ISO 42001, NIST AI) are also taken as a basis. The expert outlined the main security requirements for government agencies:
"We emphasize that AI should not have direct access to production databases and information systems of special importance. The reason is that vulnerabilities are currently emerging in AI systems, data leaks and prompt injections are occurring. This directly puts these systems at risk."
Elkhan Sadigli also proposed introducing a certain certification mechanism for the use of artificial intelligence:
"We have repeatedly discussed why not everyone is allowed to drive a car without a driver's license and why an examination is required to obtain one. Perhaps a form of license should also be introduced for the use of AI, requiring users to undergo a test (examination) before using such a device or tool."
Follow us on Telegram
t.me/apanewsaz
Follow us on Instagram
Follow us on TikTok
Follow us on Facebook
Follow us on X
