The transformation that artificial intelligence (AI) is promoting in companies has reached a new turning point with the emergence of AI agents. Given this new reality, IBM has announced an unpublished solution in the market: software that integrates for the first time the safety and governance of AI to offer a unified vision of the risks faced by organizations.
This launch combines the capacities of watsonx.governanceIBM’s governance tool, with the advanced functions of Guardium AI Securityits model for protection of models, data and use of AI. The objective: to help companies develop, deploy and supervise their AI systems, in a responsible and safe way, in line with the new global regulation.
“IA agents are called to revolutionize business productivity, but their benefits can also raise a challenge,” says Ritika Gunnar, General Manager of Data and AI in IBM. “When these autonomous systems are not governed or assured properly, they can carry important risks.”
Among the novelties that accompany this proposal, highlights the automation of key processes, such as the validation of standards against regulatory frameworks (up to 12, including the EU and ISO 42001 AI Law), the detection of unauthorized uses in distributed environments and automatic activation of governance flows.
The company launches the first platform of the sector that unifies the safety and governance of AI, in a context of growing business adoption of autonomous agents
This capacity has been reinforced thanks to the collaboration of IBM with Alltrue.AI, which provides advanced cloud monitoring functions, code repositories and integrated systems. Likewise, the tool allows us to execute Network Teaming simulations to identify vulnerabilities and define security policies that control both inputs and outputs of AI models.
“The future of AI depends on how well we ensure today. Incorporating security from the beginning is essential to protect the data, comply with normative obligations and consolidate a long -term confidence,” says Suja Viswesan, vice president of Security and Runtime Products of IBM.
Continuous evaluation and compliance from design
In addition to focusing on protection, IBM wants to guarantee responsible use of AI since design. To do this, Watsonx.Governance will allow monitoring the entire life cycle of AI agents, incorporating continuous evaluation mechanisms and traceability in aspects such as the relevance of the responses or loyalty to the context. This capacity will be key to detecting biases, errors or dysfunctions before they have an impact on the business.
Another front that addresses the solution is regulatory compliance. The new ones Compliance Accelerators They allow companies to assign preconfigured regulatory frameworks to their specific use cases, with support for United States regulations, Europe and other regions. In addition, IBM has expanded the availability of Watsonx.Governance to new data centers, such as AWS in India, which will facilitate deployment in local environments.
An answer to a growing challenge
IBM’s bet responds to a growing need for the market. The proliferation of generative AI systems and autonomous agents has generated a new level of complexity in risk management. This is highlighted by Jennifer Glenn, director of Research at IDC Security and Trust Group: “Unify the governance of AI with the security of AI provides organizations the necessary context to identify and prioritize risks, as well as the information to clearly communicate the consequences of not addressing them.”
In the middle of the career to take advantage of the potential of the agricultural AI, IBM seems determined to occupy a reference position. Its new software marks a step forward in the maturity of the business ecosystem of AI: more autonomous, yes, but also more regulated and under control.
