Microsoft
- Rank
- Score
- 5/149
- 58 out of 100
Microsoft is a US technology company founded in 1975 by Bill Gates and Paul Allen, and based in Redmond, Washington. It is known for Windows, Office and the Azure cloud. In AI it makes the Copilot assistant, which is built into many of its products, and its own MAI and Phi AI models.
Is Copilot safe?
Copilot scores 58 out of 100 and ranks 5 of 149. Its strongest category is Security, and its weakest category is Safety.
How is Copilot scored?
Copilot is scored out of 100 on Safety (25%), Transparency (25%), Security (25%), and Privacy (25%). See how we score.
Compare
Outside evidence is stronger than its written policies.
Safety framework
34Average of SaferAI's published scores.
- 62
SaferAI score: risk identification
Independent sourceSaferAI (opens in a new tab) · 15 Oct 2026 - 12
SaferAI score: risk analysis and evaluation
Independent sourceSaferAI (opens in a new tab) · 15 Oct 2026 - 28
SaferAI score: risk treatment
Independent sourceSaferAI (opens in a new tab) · 15 Oct 2026
Pre-release testing
50Points for each confirmed yes.
- Yes
Was the model tested for dangerous capabilities, such as biological, chemical or cyber risks?
Company claim50/50“This model was evaluated for safety using a combination of cybersecurity benchmarks and secure coding assessments, including CyberBench, Cyb…” MoreLess
“This model was evaluated for safety using a combination of cybersecurity benchmarks and secure coding assessments, including CyberBench, CyberSecEval, and SecRepo, to assess its ability to withstand real-world security threats, avoid introducing vulnerabilities, and align with secure coding standards.”
Microsoft model card (opens in a new tab) · 15 Oct 2026 - No
Did an outside organisation test the model before release?
Company claim0/50Microsoft model card (opens in a new tab) · 15 Oct 2026
Safety test results
88.9Average of one independent safety test of the company's models: MLCommons AILuminate. Each test is compared with the best company's result.
- 88.9
MLCommons AILuminate safety grades
Independent sourceMLCommons AILuminate (opens in a new tab) · 10 Nov 2026
Incident record
44This check counts only when a source covers the company. With no events it holds the pillar, which is not proof of a clean record. Each recorded event takes points off the pillar, and serious or recent ones count most. Allegation checks, regulator actions and open investigations use half the severity load. AI Incident Database and AIAAIC Repository, last 24 months.
15 events
AllegationAI Incident Database (opens in a new tab) · 12 Nov 2026These are reports, not proven findings.
Date Event Severity 1 Jul 2026 AIAAIC2261 · AI chatbots lure vulnerable gamblers to unlicensed betting websites1 Jul 2026 · Low (reported) Low (reported) 31 May 2026 AIID #1512 · The Sydney Morning Herald and The Age Removed an Opinion Article After Undisclosed Copilot Use31 May 2026 · Low (reported) Low (reported) 18 Nov 2025 AIID #1279 · Prominent AI Chatbots Allegedly Produced Incorrect UK Financial and ISA Guidance18 Nov 2025 · Medium (reported) Medium (reported) 24 Oct 2025 AIID #1400 · West Midlands Police Reportedly Relied on Erroneous Copilot-Generated Intelligence in Maccabi Tel Aviv Away-Fan Ban Decision24 Oct 2025 · Low (reported) Low (reported) 1 Aug 2025 AIID #1176 · Microsoft's Windows Recall Allegedly Stores Passwords and Social Security Numbers in Preview Mode1 Aug 2025 · Medium (reported) Medium (reported) 31 Jul 2025 AIID #1186 · Reported Public Exposure of Over 100,000 LLM Conversations via Share Links Indexed by Search Engines and Archived31 Jul 2025 · Medium (reported) Medium (reported) 4 Jul 2025 AIID #1218 · Microsoft 365 Copilot Vulnerability Allegedly Allowed File Access Without Audit Log Entry4 Jul 2025 · Low (reported) Low (reported) 1 Jul 2025 AIAAIC1974 · Top chatbots tricked into generating instructions on how to enrich uranium1 Jul 2025 · Low (reported) Low (reported) 1 Jul 2025 AIAAIC2116 · AI data centres spike electricity costs in Maryland, New Jersey1 Jul 2025 · Low (reported) Low (reported) 1 Jul 2025 AIAAIC2185 · Eightfold AI recruitment start-up accused of "secret" job scoring1 Jul 2025 · Low (reported) Low (reported) 25 Jun 2025 AIID #1188 · Multiple LLMs Reportedly Generated Responses Aligning with Purported CCP Censorship and Propaganda25 Jun 2025 · Medium (reported) Medium (reported) 13 Jun 2025 AIID #1106 · Chatbots Allegedly Reinforced Delusional Thinking in Several Reported Users, Leading to Real-World Harm13 Jun 2025 · High (reported) High (reported) 10 May 2025 AIID #1159 · Government‑Backed AI4Peat Mapping Tool Allegedly Misidentifies Granite Outcrops and Quarries as Peat10 May 2025 · Low (reported) Low (reported) 28 Feb 2025 AIID #956 · Alleged Inclusion of 12,000 Live API Keys in LLM Training Data Reportedly Poses Security Risks28 Feb 2025 · Low (reported) Low (reported) 26 Feb 2025 AIID #1174 · Microsoft Copilot Reportedly Able to Access Cached Data from Since-Private GitHub Repositories26 Feb 2025 · Low (reported) Low (reported)
Protecting children
0Points for each confirmed yes.
- No
Does the model card report child-safety tests or safeguards?
Company claim0/50Microsoft model card (opens in a new tab) · 15 Oct 2026 - No
Does the privacy policy set a minimum age, or say the service is not directed at children?
Company claim0/50Microsoft privacy policy (opens in a new tab) · 15 Oct 2026
Strong written policies; outside evidence is weaker.
Model documentation
75Points for each confirmed yes.
- Yes
Does the model card describe what the model can do?
Company claim25/25“MAI-Code-1.1-Flash is a text-to-text and image-to-text coding model built by Microsoft for fast, efficient assistance in everyday developer…” MoreLess
“MAI-Code-1.1-Flash is a text-to-text and image-to-text coding model built by Microsoft for fast, efficient assistance in everyday developer workflows.”
Microsoft model card (opens in a new tab) · 15 Oct 2026 - Yes
Does the model card report test results?
Company claim25/25“Quality and performance evaluations” (results shown in a table on the source page)
Microsoft model card (opens in a new tab) · 15 Oct 2026 - Yes
Does the model card state the model's limitations?
Company claim25/25“Known limitations As with any large language model, AI-generated text and code from MAI-Code-1.1-Flash may be inaccurate, incomplete, or oth…” MoreLess
“Known limitations As with any large language model, AI-generated text and code from MAI-Code-1.1-Flash may be inaccurate, incomplete, or otherwise incorrect.”
Microsoft model card (opens in a new tab) · 15 Oct 2026 - No
Does the model card describe safety safeguards?
Company claim0/25Microsoft model card (opens in a new tab) · 15 Oct 2026
Training-data disclosure
30.5Average of two independent training-data checks: the Stanford index and an academic grading of EU AI Act training-data summaries.
- 30.5
EU AI Act training-data summary grades (AI Accountability Lab)
Independent sourceAI Accountability Lab (opens in a new tab) · 10 Nov 2026
Independent transparency index
Average used40.7This outside rating doesn't cover this company, so the average of the companies it covers is used.
Transparency reports
5025 points for each area where Partnership on AI names the company a leader in reporting on real-world impacts.
- 50
Partnership on AI practice-leader areas
Company claimPartnership on AI (opens in a new tab) · 10 Nov 2026
Public safety commitments
100Average of the commitments signed.
- Yes
Has it signed the EU code of practice for general-purpose AI?
Company claim100/100“LINAGORA Microsoft Mistral AI”
European Commission (opens in a new tab) · 14 Oct 2026 - No
Did it sign only the safety and security chapter?
Company claim0/50European Commission (opens in a new tab) · 14 Oct 2026 - Yes
Has it signed the Frontier AI Safety Commitments (Seoul 2024)?
Company claim100/100“The UK and Republic of Korea governments announced that the following organisations have agreed to the Frontier AI Safety Commitments: Amazo…” MoreLess
“The UK and Republic of Korea governments announced that the following organisations have agreed to the Frontier AI Safety Commitments: Amazon Anthropic Cohere Google G42 IBM Inflection AI Meta Microsoft Mistral AI Naver OpenAI Samsung Electronics Technology Innovation Institute xAI Zhipu.ai The following organisations have been added to the existing list: Magic Minimax 01.ai NVIDIA The above organisations, in furtherance of safe and trustworthy AI, undertake to develop and deploy their frontier AI models and systems responsibly, in accordance with the following voluntary commitments, and to demonstrate how they have achieved this by publishing a safety framework focused on severe risks by the upcoming AI Summit in France.”
Frontier AI Safety Commitments (opens in a new tab) · 14 Oct 2026
No problems found in 2 public records. That is not proof.
Security programme and certifications
100Points for each confirmed yes.
- Yes
Is there a way to report security flaws, such as a bug bounty?
Company claim34/34“LinkedIn (bug bounty)”
hackerone.com (opens in a new tab) · 2 Nov 2026 - Yes
Does it state it holds a SOC 2 Type II report?
Company claim33/33“Office 365 SOC 2 Type 2 reports are relevant to system Security, Availability, Processing Integrity, Confidentiality, and Privacy.”
Microsoft security page (opens in a new tab) · 15 Oct 2026 - Yes
Does it state it holds ISO 27001 certification?
Company claim33/33“Microsoft's achievement of ISO/IEC 27001 certification points up its commitment to making good on customer promises from a business, securit…” MoreLess
“Microsoft's achievement of ISO/IEC 27001 certification points up its commitment to making good on customer promises from a business, security compliance standpoint.”
Microsoft security page (opens in a new tab) · 15 Oct 2026
Software vulnerabilities
70This check counts only when a source covers the company. With no events it holds the pillar, which is not proof of a clean record. Each recorded event takes points off the pillar, and serious or recent ones count most. Allegation checks, regulator actions and open investigations use half the severity load. Known-exploited flaws (CISA list) in the company's official software, last 24 months. A disclosed and fixed flaw does not lower the score.
No events
Independent sourceGitHub Advisory Database (opens in a new tab) · 12 Nov 2026
Data breaches
70This check counts only when a source covers the company. With no events it holds the pillar, which is not proof of a clean record. Each recorded event takes points off the pillar, and serious or recent ones count most. Allegation checks, regulator actions and open investigations use half the severity load. Confirmed data breaches, last 36 months.
No events
Independent sourceHave I Been Pwned (opens in a new tab) · 12 Nov 2026
No problems found in 1 public record. That is not proof.
Strong written policies; outside evidence is weaker.
Clear privacy policy
100Points for each confirmed yes.
- Yes
Does the privacy policy say what personal data is collected?
Company claim20/20“Your first and last name, email address, mailing address, phone number, and other similar contact details.”
Microsoft privacy policy (opens in a new tab) · 15 Oct 2026 - Yes
Does the privacy policy say whether your data is used to train AI?
Company claim20/20“As part of our efforts to improve and develop our products, we may use your data to develop and train our AI models.”
Microsoft privacy policy (opens in a new tab) · 15 Oct 2026 - Yes
Does the privacy policy say how long data is kept?
Company claim20/20“For example, when you store a document in OneDrive, we maintain the data until you actively delete it.”
Microsoft privacy policy (opens in a new tab) · 15 Oct 2026 - Yes
Does the privacy policy say who your data is shared with?
Company claim20/20“We provide data to Microsoft-controlled affiliates and subsidiaries and to vendors working on our behalf.”
Microsoft privacy policy (opens in a new tab) · 15 Oct 2026 - Yes
Does the privacy policy explain your privacy rights?
Company claim20/20“You can request access to, erasure of, and updates to your personal data.”
Microsoft privacy policy (opens in a new tab) · 15 Oct 2026
Training-data defaults
50100 if training is opt-in only, 50 if you can opt out, 0 if neither.
- No
Is your data used for training only if you opt in?
Company claim0/100Microsoft privacy policy (opens in a new tab) · 15 Oct 2026 - Yes
Can you stop your data being used for training?
Company claim50/50“You can opt out of use of your conversation activity for model training at any time.”
Microsoft privacy FAQ (opens in a new tab) · 15 Oct 2026
Export and delete your data
100Points for each confirmed yes.
- Yes
Can you get a copy of your data?
Company claim50/50“If you are signed into Microsoft Copilot with your Microsoft account, you can view, access, export and delete Copilot activity history from…” MoreLess
“If you are signed into Microsoft Copilot with your Microsoft account, you can view, access, export and delete Copilot activity history from your privacy dashboard.”
Microsoft privacy policy (opens in a new tab) · 15 Oct 2026 - Yes
Can you delete your data?
Company claim50/50“For example, you can view and clear the browsing, search, and location data associated with your Microsoft account.”
Microsoft privacy policy (opens in a new tab) · 15 Oct 2026
Privacy fines
56.9This check counts only when a source covers the company. With no events it holds the pillar, which is not proof of a clean record. Each recorded event takes points off the pillar, and serious or recent ones count most. Allegation checks, regulator actions and open investigations use half the severity load. EU privacy (GDPR) fines and Irish regulator decisions, last 5 years.
1 event
RegulatorCMS GDPR Enforcement Tracker (opens in a new tab) · 12 Nov 2026Date Event Severity 24 Oct 2024 ETid-246924 Oct 2024 · High High 2 events
RegulatorIrish Data Protection Commission (opens in a new tab) · 12 Nov 2026Date Event Severity 15 Sep 2025 DPC · Inquiry into Microsoft Ireland Operations Limited - September 202515 Sep 2025 · Low Low 15 Nov 2023 DPC · Inquiry into Microsoft Ireland Operations Limited - November 202315 Nov 2023 · Low Low
Independent privacy ratings
0Incogni's 2026 privacy-risk ranking: 100 is the lowest-risk platform rated, 0 the highest.
- 0
Incogni privacy-risk ranking
Independent sourceIncogni privacy ranking (opens in a new tab) · 10 Nov 2026
Risk governance
33Average of SaferAI's published scores.
- 33
SaferAI score: risk governance
Independent sourceSaferAI (opens in a new tab) · 15 Oct 2026
Regulator actions
62.9This check counts only when a source covers the company. With no events it holds the pillar, which is not proof of a clean record. Each recorded event takes points off the pillar, and serious or recent ones count most. Allegation checks, regulator actions and open investigations use half the severity load. UK competition regulator cases, US Justice Department settlements, last 5 years, and regulator actions and open investigations into the company's AI products at half severity, last 3 years.
1 event
RegulatorUK CMA cases (opens in a new tab) · 12 Nov 2026Date Event Severity 20 Nov 2023 Microsoft / Activision Blizzard merger inquiry20 Nov 2023 · High High 1 event
RegulatorUS Justice Department (opens in a new tab) · 12 Nov 2026Date Event Severity 12 Jun 2023 https://www.justice.gov/archives/opa/pr/microsoft-agrees-pay-20-million-civil-penalty-alleged-violations-children-s-privacy-laws · Microsoft Agrees to Pay $20 Million Civil Penalty for Alleged Violations of Children’s Privacy Laws12 Jun 2023 · Medium Medium
Court judgments
70This check counts only when a source covers the company. With no events it holds the pillar, which is not proof of a clean record. Each recorded event takes points off the pillar, and serious or recent ones count most. Allegation checks, regulator actions and open investigations use half the severity load. Court rulings against the company on privacy, consumer or competition claims, upheld on appeal, last 5 years.
No events
Regulatorcourtlistener.com (opens in a new tab) · 10 Nov 2026
Privacy contact and complaints
100Points for each confirmed yes.
- Yes
Is there a privacy contact, such as an email or a data protection officer?
Company claim50/50“If you have a privacy concern, complaint, or question for the Microsoft privacy team or Data Protection Officer, please visit our Privacy su…” MoreLess
“If you have a privacy concern, complaint, or question for the Microsoft privacy team or Data Protection Officer, please visit our Privacy support and requests page and click on “Contact the Microsoft privacy team or the Microsoft Data Protection Officer” menu.”
Microsoft privacy policy (opens in a new tab) · 15 Oct 2026 - Yes
Does it tell you that you can complain to a regulator?
Company claim50/50“You can also raise a concern or lodge a complaint with a data protection authority or other official with jurisdiction.”
Microsoft privacy policy (opens in a new tab) · 15 Oct 2026
Score history
Newest first. Only days when the score changed are listed.
3 Nov 2026
Score 58
Method 3 Nov 2026
-2
2 Nov 2026
Score 60
Method 2 Nov 2026
+4 · Method update
28 Oct 2026
Score 56
Method 28 Oct 2026, update 2
-8 · combines both 28 Oct updates (28 Oct 2026 - Update 1 and 28 Oct 2026 - Update 2)
21 Oct 2026
Score 64
Method 21 Oct 2026
-2
19 Oct 2026
Score 66
Method Not recorded
-2
15 Oct 2026
Score 68
Method Not recorded
+4 · Reason not recorded
14 Oct 2026
Score 64
Method Not recorded
-