Microsoft has announced new security tools focused on AI governance and agent safety.
The release includes 'Guardian Agents' to monitor digital assistant activity, with Microsoft Defender now capable of identifying approximately 60 types of AI assistants.
The update introduces mandatory human-in-the-loop requirements for risky automated decisions and aligns with Nvidia’s Open Agent Safety Platform to provide better visibility into agent-based workflows.
The challenge with digital assistants
The problem lies in the very nature of technology, which often operates without regard for rules or a business's judgment. As more and more such programs take on complex tasks and have access to sensitive information, Microsoft emphasizes that the ultimate responsibility for their decisions must remain with humans.
Microsoft CEO Satya Nadella has argued that control of AI systems should be independent of the companies that create them. In that context, Microsoft is participating in Nvidia's Open Agent Safety Platform, an effort to better monitor and control digital assistants.
The new protection tools
To strengthen control, Microsoft introduced the following solutions:
READ ALSO
Amazon: New TorchNeuron tool challenges Nvidia's dominance
- Guardian Agents: These are new tools designed specifically to monitor the action of digital assistants.
- Upgraded Microsoft Defender: The company's security program can now detect about 60 different types of digital assistants used by employees.
- Human approval: In cases where an action by the digital assistant is deemed dangerous, the system will now require human approval to proceed.
In addition, the company has released its Surface Laptop Ultra computers, which have built-in control systems specifically for such assistants.
G
Hacks.gr on Google
Set it as a preferred source for cybersecurity updates.
Preferred source
Top comments (0)