Digital platforms

Microsoft Corp.

Rank: 1st
Score: 50%

Headquartered in the United States, Microsoft offers software, hardware, cloud storage, search, email, chat applications, and the social networking service LinkedIn. The latter has over 1 billion members, while the company’s search engine has 100 million daily users. It derives most of its revenue from its cloud-based solutions, as well as its Office Suite, Windows operating system, and advertising on its digital platforms.

Meta3
47%
Apple4
44%
Kakao4
44%
X7
40%
Yandex8
37%
Baidu9
33%
Tencent11
30%
Samsung12
28%
Amazon13
27%
VK13
27%

Microsoft has emerged as a major player in AI, investing nearly USD 14 billion in OpenAI as of 2024 and announcing a plan to invest USD 80 billion in AI-enabled data centers in 2025. The company has also continued rolling out AI features in its consumer products. For instance, it has added generative AI search features to Bing and introduced AI-powered features for its premium users on LinkedIn, including a chatbot that provides job search assistance. In 2024, LinkedIn automatically began using members’ accounts and data for the training of its AI models without their prior consent. In January 2025, a lawsuit filed in the U.S. accused the platform of sharing users’ private messages with other companies to train AI models. It also deepened its collaboration with the Israeli government, providing the Israeli military with “large-scale access” to OpenAI’s GPT-4 model as well as access to its Azure cloud computing platform, to support its bombardment of Gaza.

As part of its effort to integrate its messaging and internet calling services, Microsoft announced that it would shut down Skype—once the most popular internet-based phone and video service in the early 2000s—for individual users starting in May 2025. As a result of this change, the 2025 Index will be the last time that Skype is included in our assessment of Microsoft. The company has said that Skype users will have the option to be moved automatically to Microsoft Teams, along with their user data.

In late 2023, Microsoft shareholders rejected a proposal by impact investment firm Arjuna Capital asking that the company share a report assessing misinformation and disinformation risks associated with generative AI and how it will “remediate those potential harms.” Earlier that same year, Microsoft notably laid off its entire AI ethics team. In 2024, the company did not respond to questions submitted by the Business and Human Rights Resource Centre about its “human rights due diligence efforts when investing in generative AI startups and companies.” However, LinkedIn and Bing began conducting systemic risk assessments, as required for very large online platforms (VLOPs) under the European Union's Digital Services Act, publishing their first reports in August 2023.[1]

Despite Microsoft ranking first, its performance largely stagnated. The company made a commitment to respecting human rights in its deployment and development of generative AI that failed to account for all other AI systems it deploys or develops. The company also continued to disclose little information about its human rights due diligence processes for the development and deployment of algorithmic systems. It did, however, explain how it uses algorithmic systems to curate, recommend, and/or rank content on Bing and LinkedIn, and disclosed options for users to control how content is recommended to them.

Key takeaways

  • While Microsoft was transparent about its due diligence processes in relation to government regulations, it did not disclose whether or not it conducted human rights impact assessments on its processes for terms of service enforcement or on its targeted advertising policies and practices. It disclosed only limited information about human rights impacts associated with its use and development of algorithmic systems.
  • Microsoft’s advertising content rules lacked clarity, particularly for LinkedIn. It was even less transparent about its advertising targeting rules, failing to share any information about the types of targeting parameters that are forbidden for all of its services except Bing, and disclosing nothing at all on how it enforces these rules.
  • The company was transparent about which user information it collects, but it disclosed only some of the information it infers and shares. It did not provide clear retention periods for the information it collects, nor did it disclose sufficient options for users to control and access their information.

Key recommendations

  • Improve human rights due diligence processes. Microsoft should conduct robust and regular human rights impact assessments to assess and mitigate risks associated with its processes for policy enforcement, targeted advertising policies and practices, and its deployment and development of algorithmic systems.
  • Provide users with more control. Microsoft should publish more information about users’ options to access and control how their data is collected and used.
  • Be more transparent about third-party demands. The company should be more transparent in its process for handling government censorship demands for LinkedIn. Additionally, it should clarify its handling of private requests for content removal and user information.

Services evaluated:

  • Bing
  • Outlook.com
  • Skype
  • OneDrive
  • LinkedIn

The 2025 RDR Index: Big Tech Edition covers policies that were active on August 1, 2024. Policies that came into effect after August 1, 2024, were not evaluated for this benchmark.

Scores reflect the average score across the services we evaluated, with each service weighted equally.

  • Lead researchers: Afef Abrougui, Mila Bajic

Changes since 2022

  • Microsoft no longer disclosed whether it assesses the privacy and freedom of expression impacts of its terms of use.
  • Microsoft disclosed more about its processes for enforcing its terms of service, including its use of algorithmic systems to detect infringing content.
  • The company no longer provided data on the number of advertisements it restricted to enforce its Bing advertising policies.
  • Microsoft became more transparent about its handling of data breaches by disclosing its process for notifying data subjects who might be affected by a data breach and the kinds of steps it will take to address the impact of a data breach on its users.

Scores since 2017

100%0%20172018201920202022202562%61%62%50%50%50%
Most companies’ scores dropped between 2019 and 2020 with the inclusion of our new indicators on targeted advertising and algorithmic systems. To learn more, please visit our Methodology development archive.
Governance59%
Freedom of expression44%
Privacy52%

We rank companies on their governance, and on their policies and practices affecting freedom of expression and privacy.

Governance 59%

Microsoft ranked second in this category, behind Meta. The company disclosed a strong and explicit commitment to human rights in accordance with international frameworks, but it failed to extend that commitment to all the types of AI systems it deploys and develops (G1). It had senior leadership oversight over how company policies and practices affect freedom of expression and information, as well as privacy (G2). It also had an employee whistleblower program and training program covering human rights issues (G3). The company disclosed that it conducts robust human rights impact assessments on how government regulations affect freedom of expression and privacy on its platforms (G4a). However, it fared poorly on due diligence processes in relation to the human rights impacts of its processes for policy enforcement, targeted advertising policies and practices, and the development and deployment of algorithmic systems (G4b, c, d).

Freedom of expression 44%

Microsoft ranked fourth in this category, behind Bytedance, X, and Alphabet. While its score improved by four percentage points in this category, it still lacked clarity about its advertising content and targeting policies and how it enforces them (F3b,c). Further, it stopped publishing data about the number of advertisements it restricts for violating its rules (F4c). The company revealed some information about how it handles private content removal requests, such as copyright requests for Bing and LinkedIn, but it offered no insight into this process for Skype and OneDrive (F5a). It also did not make a commitment to exercise due diligence or push back on overbroad requests received for LinkedIn.

Privacy 52%

Microsoft ranked second, behind Apple, in the privacy category. It clearly disclosed which user information it collects (P3a) and some of the information it infers (P3b) and shares (P4). However, it lacked clarity about the purposes for collecting, inferring, and sharing this information (P5) as well as its data retention policies (P6). It had clear, rights-respecting policies for handling government demands for user information (P10a). However, it failed to commit to carrying out due diligence on requests made through private processes or to push back on inappropriate or overbroad requests (P10b). The company had relatively strong disclosures about some of its security policies, disclosing that it has systems in place to limit employee access to user information and sharing that it conducts internal and third-party audits on its products and services (P13). It also explained its process for notifying data subjects of data breaches and the kinds of steps it will take to address the impact of a data breach (P15). However, while it disclosed that it encrypts the transmission of user communications by default, it did not clarify if each transmission is encrypted using unique keys (P16).

Indicators

Footnotes

[1] The LinkedIn and Bing reports reports were not publicly accessible until the EU released the risk assessment reports submitted by VLOPs in November 2024. As they were published after the policy cut-off date, they were not considered in this assessment.