Azerbaijan is preparing additional requirements for government agencies using third-party artificial intelligence systems, including products from foreign developers, to prevent official information from leaking.
Nadir Novruzov, a department head at the State Service for Special Communication and Information Security, announced the plans at a summit of government IT leaders in Baku on Oct. 20, Report said.
Agencies will be sent model provisions to add to their information security instructions. These will set out how employees should use external AI services without exposing official data. Separate security requirements are being developed for AI systems built within government agencies.
AI Agents Should Have Limited Access
Novruzov also warned against giving AI agents the usernames and passwords of live databases and information systems. Some developers and system administrators already do so, he said.
He urged agencies to manage deployed AI systems and the tools they use, and to give agents only the minimum access privileges needed for a task. An error by a system with extensive permissions could have serious consequences, he warned.
The security requirements are expected to be adopted soon, Novruzov said, without giving a specific date. Work on ethical rules for AI use is expected to follow.
He added that the Azerbaijan Standardization Institute, AZSTAND, had already adopted five AI standards, while the Cabinet had identified 11 priority areas for 2026.
Source: Vesti Baku.