As regards Microsoft Ai Tour By the end of last year, we reached Brussels, Paris and Berlin and gave European organizations that were under tension of the possibility of our latest AI technologies and participated in the deployment projects. They were also warned that the year 2025 is the year when key obligations under the AI European Union will come into force and open a new chapter in digital regulation because the First World Law is becoming a reality.
At Microsoft, we are ready to help our customers do two things at once: to innovate with AI and observe the EU AI law. We are building our products and services to comply with or obligations under the EUN law and cooperate with our customers to help them deploy and use the technology compatible. We also undertake with European policy creators to support the development of efficient and efficient implementing procedures within the EU act, which is in line with developing international standards.
Below we go in more detail about this effort. The sale of data for compliance with the EU AI law is distributed, and the key details of the implementation are not yet completed, we will publish information and tools on an on boi. You can consult our ACT documentation on Microsoft Trust Center stay up to date.
MICROSOFT products and services that adhere to the EU law
Organizations surround the world that uses Microsoft products and services for the innovative AI solutions that emphasize them to achieve more. For these customers, especially those who work worldwide and across different jurisdictions, they have regulation compliance. That is why, in every Agreement on Microsoft customers, all the laws and regulations relating to Microsoft have committed to complying. This included the EU AI law. Therefore, we also made timely decisions to build and continue investing in our AI management program.
As shown in our opening Transparency ReportWe have accepted the approval of risk management, which covers the entire AI development cycle. We use practices such as impact assessment and red teams to help us identify potential risks and ensure that teams of building models and systems with the highest risk receive further supervision and support through public management processes, such as our Sensitive Program. After risk mapping, we use systematic measurements to evaluate the prevalence and the severity of the risks against defined metrics. Risks are managed by implementation of mitigation, such as classifiers that are part of Azure AI content safety and ensuring continuous monitoring and response to incidents.
Our framework for the management team of Microsoft AI Solutions – AI responsible standard– After it was developed with regard to the early version of the EU law.
Based on these endowment components of our program, we devoted significant sources to the implementation of EU AI events across Microsoft. Functional working groups combining AI administration, engineering, legal and experts in public policy have been working on identifying for months as to whether and how they should be updated at our internal stalls and practices to detect the final text of the implementation of details. They also identified any other engineering work needed to ensure readiness.
For example, the provisions on prohibited EU ACS practices are among the first provisions to enter into force in February 2025. We have shown other instructions before the newly established AI of the European Commission providing other instructions. This included:
- Performing a thorough review of Microsoft -owned systems Already in the market, identify all places where we might need to modify our consent, including updating or implementation of technical mitigation. Do that, We have developed a number of questions designed to raise whether the AI system could implement a prohibited practice and through our central tool we left this survey for engineering teams. The issued experts reviewed the answers and followed the teams directly, where further clarity or other steps were required. These screening questions remain in our central responsible workflow Workflow tool to keep teams working on new AI systems and engage the workflow as needed.
- Creating new limited uses In our internal corporate policy to ensure that Microsoft does not propose or deploy AI systems for prohibited EU law. We are also developing specific marketing and sales instructions to ensure that our general AI technologies are not launched or sold for use that could impose prohibited procedures of the EU AI law.
- Update of contracts, Including our General Code of Behavior AITo make our customers clearly understand that they cannot participate in the forbidden practices.
We were also among the first organizations to apply for three main obligations in AI PactA set of voluntary promises of developed offices and to support regulatory readiness before some of the upcoming deadlines for compliance with the EU AI Act. In addition to our regular rhythm of the publication of annual responsible AI transparency reports, you will find an overview of our EU AI law and a more detailed summary of how we make the provisions on prohibited practices in Microsoft Trust Center.
Customer cooperation to help them deploy and use Microsoft products and services in accordance with the EU law
One of the main concepts of the EU AI law is that obligations need to be allocated across the AI supply chain. This means that a regulated current actor, such as Microsoft in its function as a tool, service and component provider, must support regulated Downstream actors such as our business customers when they integrate the Microsoft tool into a high Ri-Is IS system. We accept this concept of shared responsibilities and like to support our customers with their development and deployment AI activated sharing our knowledge, providing documentation and offering tools. It all the ladders until Customer obligations AI We created in June last year to support our customers on their responsible AI.
We will continue to publish documentation and resources related to the EU AI law in Microsoft Trust Center to provide updates and address customer questions. Our Responsible Web AI Sources It is also a rich source of tools, practices, templates, and information that Mayy of Mayym from our customers will help create the basics of good public administration to support adherence to EU AI laws.
The leading documentation 33 transparency notes that we have published the Honest Year 2019, providing basic information about the abilities and limitation of our tools, components and services AI, which our customers rely on as users of the Microsoft AI platform. We also published documentation for our AI systems, such as Aswers, which often ask questions. Our Transparency Note for Azure Openai servicePlatform AI and FAQ for What -the -pilotAI system is examples of our approach.
We expect several secondary regulatory efforts under the EU law AI to provide additional instructions for model and system documentation. These standards for documentation and transparency are still ripening and benefit from another definition, lies in efforts such as Report For the Hiroshima AI Code of Code of Code for organizations developing advanced AI systems. Microsoft has been pleased with the intelligence frame through the process of ConventD by OECD and is looking forward to its solid public edition.
Finlly, Becuse Toolling is necessary to achieve consistency and effective compliance, we have made our customers accessible to our customers that we use for our own internal purposes. These tools include ComplianceWhich helps customers understand and take steps to improve the ability to comply with regulations in many regulatory areas, included the EU AI law; Azure AI content safety Help relieve content -based damage; Azure AI Foundry Help with the evaluation of generative applications AI; and A tool identification of python or pyrite risk identificationThe open innovation framework that our independent AI Red team sets to help identify the potential damage associated with our models and systems with the highest RI.
Helping the development of efficient, efficient and interoperable implementation procedures
The unique feature of the EU AI Act is that there are more than 60 secondary regulatory efforts that will have a significant impact on defining the exclusion of difficulties and directing the organization. Since this effort is underway or is still underway, we are in a key window of the opportunity to help determine the implementation that is effective, efficient and aligned with the emerging international standards.
Microsoft is engaged in the EU’s central regulatory body, AI and other relevant bodies in the EU Member States to share the knowledge of our development, administration and compliance, sought the clarity of open issues and advocates practical results. We also participate in the development of the Code of Practice for the General-Purpose AI model and remain long-term contributors to the technical standards developed by European standards such as prices and CENELEC, to address the high-risk AI in AI system in the EU AI.
Our customers also have to play a key role in this implementation effort. Involvement in the creators of politics and industrial groups to understand the developing requirements and have a word about them, our customers have the opportunity to contribute their valuable knowledge and to help shape executions that better reflect their circumstances and needs, and recognize a wide range of organizations that are, that are open to the opportunity to innovate and grow with AI. In the coming months, the key question is solved when an organization that is substantially fine -tuned by AI models becomes Downnstream providers to meet general AI model obligations in August.
Forward
Microsoft will continue to produce significant products, tools and investment in administration to help our customers innovate with AI in accordance with new laws such as EU AI. Implementation procedures that are effective, efficient and interoperable internationally will be the key to supporting useful and trustworthy innovations on a global scale, so we will continue to lean towards regulatory processes in Europe and worldwide. We are excited to see projects that revive our companies Microsoft on tour in Brussels, Paris and Berlin, improve people’s lives and gain their confidence and welcome feedback on how we can continue to support our customers in their efforts to comply with new laws as law EU AA.