Microsoft has signed a legally binding agreement with the American Federation of Teachers (AFT) that prohibits AI companions or features "designed to foster emotional attachment or dependency" and bars the use of student or educator data to train its AI systems. The deal, announced on 15 September 2026, arrives as public backlash against AI in classrooms intensifies in both the United States and Europe.
Key provisions of the Microsoft‑AFT pact
The Euronews report confirms that the standards are enforceable and require third‑party audits to verify compliance. Microsoft also commits that data it collects in schools cannot be sold, used for advertising, or employed for product development, with a narrow exception for data that could help protect students. Brad Smith, Microsoft’s vice‑chair and president, said the framework "sets a high bar for child privacy and AI safety" and that the company intends to extend it to every school district nationwide.
“This standard sets a high bar for child privacy and AI safety, and we’ll extend this agreement to every school district across the country,” – Brad Smith, Microsoft vice‑chair and president.
Josh Golin, executive director of the online‑safety nonprofit Fairplay, was quoted asking whether Google will adopt a similar framework, noting the question remains unanswered.
“Is Google going to embrace this framework? That is a huge question,” – Josh Golin, Fairplay.
Impact on schools and educators
Under the agreement, any AI tool that attempts to keep a student engaged beyond the learning task – for example, a chatbot that encourages prolonged conversation – must be disabled in classrooms. The prohibition also covers any feature that could create a dependency relationship, a concern raised by teachers who fear AI could replace human interaction.
Data‑use restrictions mean that Microsoft will not feed student‑generated text, voice recordings, or usage logs into its large‑language‑model training pipelines, except where the data is needed to protect a child from harm. Families will receive plain‑language explanations of how the tools work and what data is collected, a transparency measure that the agreement requires.
Google’s silence
The Euronews article notes that Google, the dominant U.S. ed‑tech provider, has not said whether it will adopt comparable protections. No public statement from Google or its spokesperson was found in the source material, leaving schools that rely on Google Workspace for Education without a clear policy on AI companion bans or data‑training restrictions.
Microsoft’s scale and financial backdrop
Microsoft remains one of the world’s largest technology firms. According to its most recent Form 10‑K filed on 29 July 2026, the company reported net income of $133.749 billion for the fiscal year ended 30 June 2026 and total assets of $758.376 billion at the same date. Shareholders’ equity stood at $442.387 billion, and the company had 7.427 billion shares outstanding.
For context, the filing also includes a 2010‑Q3 revenue figure of $36.148 billion, illustrating the company’s growth over more than a decade. While the agreement does not involve a monetary transaction, the financial heft of Microsoft underscores the significance of its voluntary, audit‑backed commitments.
| Metric | Value | Period | Form |
|---|---|---|---|
| Revenue | USD 36.148 billion | 2010‑07‑01 to 2010‑12‑31 | 10‑Q |
| Net income | USD 133.749 billion | 2025‑07‑01 to 2026‑06‑30 | 10‑K |
| Total assets | USD 758.376 billion | 2026‑06‑30 | 10‑K |
| Source: Microsoft SEC filings (Form 10‑Q 2011‑01‑27; Form 10‑K 2026‑07‑29) | |||
What remains unknown
The agreement’s legal language is public, but the exact audit schedule and the identity of the third‑party auditors have not been disclosed. Microsoft also left a narrow exception for data that could help protect students, but the criteria for invoking that exception are not detailed in the source.
Google’s future stance is another open question. The Euronews piece reports no comment from the company, and no other source in the packet provides a timeline for any potential policy announcement.
Looking ahead
With the agreement effective immediately after signing on 8 September 2026, schools that adopt Microsoft’s AI tools must adjust their curricula and procurement processes to comply with the new guardrails. The requirement for third‑party audits suggests ongoing monitoring, which could become a benchmark for other ed‑tech vendors.
Should Google eventually issue a comparable framework, the market could see a de‑facto industry standard for AI safety in education. Until then, educators and parents will be watching how Microsoft’s commitments translate into classroom practice and whether the audit reports reveal any compliance gaps.