Kami Platform Evaluation Report – May 2025
TL;DR Executive Summary
Kami is a mature, widely adopted K–12 platform with strong privacy practices, teacher-centered AI tools, and broad district-level support. It offers optional AI features (e.g., summarization, auto-grading, re-leveling) with full teacher oversight and administrator control. Kami stores minimal student data, complies with FERPA/COPPA/SOPIPA/state laws, and allows districts to fully manage access. It has no known data incidents, and multiple case studies show positive educational outcomes. Recommendation: 🟢 Go — Kami is safe and effective for classroom deployment at St. Joseph Public Schools.
Table of Contents
- AI Functionality & Purpose
- Data Handling & Privacy
- Safeguards by Age Group
- Regulatory & Legal Compliance
- AI Safety & Oversight
- District-Level Implementations & Reviews
- Transparency & Governance
- Evaluation Matrix
- Final Recommendation
- Key References and Resources
1. AI Functionality & Purpose
Core AI Features: Kami has introduced several AI-powered tools to enhance teaching and learning. Key features include Questions AI (an autograding question generator), content summarization, text “re-leveling” (reading level adjustment), explanation of highlighted text, and translation tools. These “Understand” features aim to help differentiate instruction – for example, summarizing complex readings into simpler outlines or rephrasing text for clarity. Kami also offers AI-assisted assessment creation: teachers can auto-generate quiz questions (with automatic grading) from existing documents or curriculum content. Additionally, AI is used to provide insights in the Class View dashboard (real-time monitoring of student work) and to support personalized feedback workflows.
Introduction Timeline & Outcomes: Kami’s AI capabilities are relatively new. The Questions AI tool was rolled out around late 2023, with refinements through 2024 (help articles were updated ~September 2024). By early 2025, Kami expanded its AI toolkit with the Summarize, Explain, Translate, and Relevel tools as part of an “Understand” feature set. In April 2025, Kami’s updates highlighted these tools as “game-changers” for comprehension and accessibility. The intended outcomes are to save teachers time (e.g. auto-generating and grading quiz questions in seconds) and to enhance student understanding (e.g. quickly condensing a dense text or adapting it to a student’s reading level). The AI features are primarily assistive– helping create assessments, summaries, or translations – rather than student-facing chatbots. This aligns with Kami’s goal of helping teachers manage classrooms, create assessments, grade assignments, and adapt materials to student needs.
Optional vs. Always-On: Importantly, Kami’s AI features are optional and require explicit opt-in. By default, teachers/admins must enable the AI tools before use. For example, to use Questions AI, a teacher must opt in to “Kami AI” in their account settings. Schools can choose if and when to activate these AI functionalities; they are not forced on by the system. Even after enabling, teachers maintain control – e.g. they can generate quiz questions with AI, then edit or discard any AI-generated content before sharing with students. This opt-in design ensures that AI is used deliberately as a tool, not as an automatic or always-on component. (Looking ahead, Kami has hinted at a “Kami Companion” AI assistant launching in mid–2025, which will likely also be an optional feature that districts can evaluate separately.)
2. Data Handling & Privacy
Student Data Collected: Kami collects only limited personal information from students, primarily account information (name, school email, etc.) needed to use the app. It does not collect sensitive personal details beyond what the school or teacher provides when rostering students. The content students create (annotations, responses, etc.) is considered student data but, notably, Kami does not typically store the original documents themselves on its servers. Instead, student work is saved to the student’s or school’s own storage (e.g. Google Drive or OneDrive), with Kami storing only the annotations and metadata needed to sync and display that work. For example, when a PDF assignment is opened in Kami, the file stays in Google Drive; Kami’s cloud saves the highlights, comments, or answers the student adds, and re-applies them when the file is reopened. This dual system (Kami Cloud + Drive/OneDrive) means documents reside in the school’s cloud, and Kami’s servers hold just the layer of annotations (and even that is synced to Drive if integrated). An exception is if a teacher deliberately uses Kami’s “Share Document” feature to upload a file for distribution – in that case, the file is temporarily stored in Kami’s cloud for sharing, but can be removed by the teacher at any time.
Data Inferred or Generated by AI: When using AI features (e.g. generating questions or summaries), Kami’s system will process the content of a document or prompt to produce the result. Kami has not published detailed technical disclosures of these AI processes, but it’s likely that third-party large language models(such as OpenAI or similar services via Microsoft/Google) are used under the hood for text generation. Kami’s privacy principles indicate that any third-party services used have committed to Kami’s data protection standards. In practice, this likely means any data sent to an AI service is protected and not used for other purposes. For instance, if a teacher uses Questions AI on a PDF, the textual content might be sent securely to an AI engine to generate quiz questions, with no human review on the provider side. Kami does not use any student data for advertising or profiling, and generated AI content is not used to build student profiles – it’s ephemeral output for instructional use.
Storage & Geographic Location: Kami utilizes reputable cloud infrastructure. According to the company, their servers run on Amazon Web Services (AWS) and Google Cloud Platform (GCP). These data centers provide encryption and security at scale, and Kami notes that its data security posture leverages AWS/GCP best practices. For U.S. schools, student data is stored on servers located in the United States whenever required by law or district policy. In fact, standard data privacy agreements with schools stipulate that student records will be stored in the U.S. and that Kami will disclose data storage locations upon request. All data in transit is encrypted (HTTPS) and data at rest in Kami’s cloud is also encrypted. This includes annotation data, account info, and any files temporarily uploaded – everything is encrypted both in transit and at rest.
Third-Party Integrations (AI & Analytics): Kami integrates with school platforms (Google Classroom, Canvas, Schoology, Microsoft) for single sign-on and file storage, but does not embed third-party trackers for ads or analytics on its site. They explicitly state they do not host any third-party cookies or trackersthat would profile users. Usage analytics are presumably collected in-house (for example, to show a teacher the student usage stats or to improve the product), but there is no evidence of invasive tracking. For AI services: while not named outright in public documentation, it’s reasonable to assume Kami’s new AI features rely on external AI platforms (such as OpenAI’s GPT or similar models via cloud APIs). Kami likely routes data to those AI engines securely and does not retain the AI queries beyond providing the result(nor allow the AI provider to use it for training, in line with education data privacy commitments). Additionally, Kami’s privacy policy mentions using only third-party services “that have made the same commitment to us”regarding data use, suggesting any AI or cloud vendors must agree not to misuse student data.
Anonymization & Retention: Kami’s policies ensure that student work and personal info are used solely to provide the service. Data is not sold or shared for marketing. If any diagnostic or product improvement analysis is done, it would be on anonymized or aggregate data (Kami does not explicitly say this, but their stance on privacy suggests they would strip personal identifiers if analyzing usage trends internally). Data retention is governed by the school’s needs: schools retain ownership of student records, and Kami will delete data upon request or when a contract ends. In fact, in various agreements Kami commits to permanently delete all customer data within 90 days of contract termination (or sooner upon request). Moreover, any user (or parent) can request data deletion at any time – Kami’s privacy policy states “if you ever want to be forgotten, simply email us… and all your data will be completely removed”. This manual opt-out gives families an avenue to have data wiped if needed.
Opt-In/Out Controls: Beyond the AI opt-in already noted, Kami provides granular admin controls for data access. For instance, district admins can limit third-party app access via Google/Microsoft scopes (Kami publishes guides on managing OAuth permissions). Administrators can also disable specific features on a per-role or per-user basis via Kami’s Feature Permissions dashboard. This means a district could, for example, turn off the new AI “Understand” tools for all students (or just for certain grade levels) while leaving them enabled for teachers. Schools can thus opt out younger students from using AI features if desired, or generally tailor which Kami features are available to which users. These safeguards allow compliance with parental opt-out requests as well – e.g. if a parent doesn’t want their child using a particular Kami feature, the school could disable that feature for the student’s account.
In summary, Kami’s data practices are very education-privacy-centric. Student data collection is minimal and for educational purposes only, data is kept on secure cloud servers (AWS/GCP in the U.S.), and the company contractually commits to strict limits on data use (no selling, no advertising, deletion on request, etc.). These practices help Kami comply with federal and state privacy laws (discussed next) and align with district expectations in Michigan.
3. Safeguards by Age Group (K–12)
Differentiated Protections: Kami is used across elementary, middle, and high school, and while the core application is the same, the admin controls allow differentiation by age group. Rather than offering separate child vs teen versions, Kami provides feature-level permissions so schools can adjust what younger students can access. For example, a district might disable the AI-powered “Understand” tools (Summarize/Explain/Relevel) for elementary students until they are older, or turn off certain collaboration features for younger grades. Through the License Dashboard, admins can toggle features on/off by user role (Teacher, Student, Other) and even for individual users. This effectively means elementary students can have a more restricted Kami experience (fewer features, if deemed appropriate), while high schoolers might have full access. All students under 13 are required to use Kami under a school account (with school-provided consent under COPPA) – Kami’s Terms specify that users under 13 should only use the service with a teacher or parent authorizing it. Thus, elementary usage is always within the school-managed context, ensuring COPPA compliance and adult oversight.
Content Moderation & Filters: Kami’s design minimizes open-ended inputs from students to the AI, which reduces the risk of inappropriate content. There is no public chat or free-form AI bot for students. The AI features students might use (translate, summarize, read-aloud) are constrained to educational content provided by the teacher. For instance, a student can highlight a passage from their assignment and click “Summarize,” but they cannot ask the AI arbitrary off-topic questions. This reduces exposure to unsupervised AI output. Moreover, any AI generation is subject to the moderation filters of the underlying AI model (e.g., if using OpenAI’s API, it has built-in content filters). While Kami hasn’t published a specific “bad word” filter list, it’s reasonable to assume that common profanity or unsafe content would be filtered out by the AI provider before reaching the student. Kami’s Acceptable Use Policy also prohibits users from inputting harassing, violent, or pornographic material into the platform; presumably, if a student tried, the AI tools would refuse to process it or the teacher would see it and intervene.
Teacher Visibility and Controls: One of Kami’s strongest safeguards is that teachers have real-time visibility into student work. With the Class View feature, a teacher can see thumbnails of each student’s document and watch annotations appear live. If a student were to write something inappropriate or off-task, the teacher would notice immediately. Teachers can also jump into a student’s document live to guide them. This acts as a deterrent against misuse and a safety net – especially for younger students who may need closer monitoring. Teachers can override or erase any student annotation if needed (for example, remove an inappropriate comment), maintaining classroom control.
Escalation & Reporting: For serious issues (e.g. a student types a threat or something concerning in a Kami document), Kami itself doesn’t appear to have an automated escalation to admins (no AI scanning student content for self-harm keywords, etc., as of now). However, since the student’s content belongs to the district and is stored on district-controlled drives, normal school monitoring policies apply. A district could use Google’s tools or other DLP systems on Drive to catch flagged content, if they have those in place. Kami’s own role is neutral in this sense – it doesn’t insert its own surveillance, respecting student privacy (no data mining), but also doesn’t actively police content beyond basic use restrictions. Any observed misconduct can be addressed by the teacher in real time or reported to administration through normal disciplinary channels.
Age-Appropriate Experience: Kami’s interface includes various tools like Text-to-Speech and Dictionary, which are very useful for younger learners or those with reading difficulties. These are not AI in the controversial sense (they’re standard assistive tech), but they illustrate that Kami tailors the experience to different needs (special education and early readers get support). For AI features, the “Explain” tool could help younger students by defining words or simplifying a sentence – effectively acting as a built-in tutor, but again under teacher guidance. If a student at any level encounters an AI-generated output that is confusing or seems off, they are encouraged to ask their teacher; the teacher can always override AI content (e.g., edit a quiz question that the AI created poorly).
In summary, Kami provides universal safeguards (no ads, no external chat, teacher oversight) that protect all students, and additional administrative controls to scale permissions appropriately by grade level. Elementary students’ data is protected by requiring school consent and limiting features; older students might use more of the AI tools, but always within the safe “walled garden” of the classroom workflow. This approach meets COPPA obligations (verifiable consent via the school) and supports differentiated usage policies for different age groups as needed.
4. Regulatory & Legal Compliance
FERPA: Kami affirms that it complies with the Family Educational Rights and Privacy Act. As a “school official” under FERPA (when under contract with a district), Kami only uses student data for educational purposes authorized by the school. It does not redisclose education records except as permitted (e.g. to a sub-processor like AWS under strict privacy terms). In practice, FERPA compliance is seen in Kami’s contract terms: districts retain ownership/control of all student records, and Kami agrees to use them only to provide the service. Kami also provides mechanisms for parents or eligible students to review and correct information via the school (for example, a parent can request account info or deletion via the district or directly through Kami’s support) – meeting FERPA’s access requirements. The Common Sense Privacy evaluation gave Kami a 93% score for FERPA compliance, indicating a high likelihood that all FERPA obligations are met.
COPPA: For users under 13, Kami’s policy is to obtain consent via the school – which is allowed under COPPA for educational products. Kami is explicit about COPPA compliance in its privacy policy and documentation, and it does not collect more info than necessary from children. There are no ads or behavioral tracking (so no COPPA-prohibited profiling). Account creation for students under 13 is done through Google Classroom or similar SSO, with the school handling parental consent as needed. Kami’s terms also state that by using the service, you confirm you are either (a) over 13, or (b) under 13 with school/parent permission. This check places the onus on schools to ensure compliance, which is standard. The company’s strong stance of not sharing data with third parties beyond service providers also supports COPPA – there’s no unauthorized disclosure of kid’s info. Common Sense’s review gave Kami a 93% score for COPPA as well.
SOPIPA (California) and Other State Laws: Kami adheres to state student privacy laws like California’s SOPIPA and similar statutes across the US. In fact, Kami’s own security agreements (from California districts) list out the SOPIPA requirements point by point, and Kami agrees to all – e.g. no targeted advertising, no selling student data, no creating student profiles except for school purposes, robust security and deletion policies, etc.. They check all those boxes with “Yes” in their compliance documents. Additionally, Kami’s privacy pledge references “all USA state laws on data privacy” – which would include Michigan’s laws. Michigan’s Student Online Personal Protection Act (if referencing SOPIPA-like protections) and the Michigan Mental Health Code amendments for educational data are all about not selling data and keeping it secure. Kami’s commitments (no selling, ownership remains with schools, etc.) align with those requirements. The company has likely signed Michigan’s statewide Data Privacy Agreement via the Student Data Privacy Consortium (SDPC) if used by Michigan districts. (Michigan is part of the SDPC framework; while we don’t have a specific Michigan DPA cited, Kami has signed standard NDPA contracts in many states – for example, an Oklahoma district agreement in 2023 and a Louisiana state contract in 2023 – which include all the typical legal safeguards and would satisfy Michigan law as well.)
Student Privacy Pledge: Kami is a signatory of the (now-retired) Student Privacy Pledge. By signing that pledge, Kami publicly committed to the 12 key principles of K–12 data protection: not selling student info, not using data for behavioral ads, using data for authorized education purposes only, supporting parental access, security standards, breach notification, etc. (The pledge was retired in April 2025 with the introduction of an updated framework, but Kami’s inclusion on the list of former signatories demonstrates a long-term commitment to those principles.) Furthermore, Kami earned a “Privacy Verified” seal from Common Sense Education – indicating its policies were vetted to align with best practices and legal requirements.
SDPC and DPAs: Kami is willing to sign Data Privacy Agreements (DPAs) with districts. It has executed the National Data Privacy Agreement (NDPA) version 1.0 with districts in multiple states (for example, Moore Public Schools in OK, signed Feb 2023). These model agreements contractually bind Kami to comply with FERPA, COPPA, state laws, and specify data handling rules (e.g. data storage in US, breach notification within X days, deletion timelines). Michigan districts can leverage the SDPC framework to get a signed DPA from Kami, as other districts have done. Kami also often provides a Security & Data Protection documentation package (as was done for a California district in 2019) detailing its compliance with laws like AB1584 (student records protections) and outlining security controls. In one such document, Kami confirmed that “student content is saved to their own storage (Google Drive or local)”, that the district retains ownership of data, and that Kami cannot disclose or sell student information beyond the school’s purposes.
GDPR and International Compliance: Though the focus here is US, it’s worth noting Kami also complies with GDPR for any EU data (they have users globally) and generally meets high international standards for privacy. They likely act as a Data Processor under GDPR, only processing data on instructions from the school (Data Controller). Their General Data Privacy Policy aligns with GDPR principles like data minimization and access rights.
Breach Notification and Security Audits: Legally, Kami is prepared for worst-case scenarios too. The privacy policy states that in the event of a security breach, Kami will notify affected schools/users within 7 days via website and email, which meets the stringent breach laws (e.g., many states require notice within 30 days, Kami’s promise is even faster). Contracts (like the Louisiana state DSA) also require Kami to cooperate with any security audits or investigations into compliance – and Kami agrees to that. To date, there have been no publicized data breaches involving Kami.
In summary, Kami demonstrates strong compliance across the board: FERPA, COPPA, SOPIPA, PPRA, state-specific acts, etc. are addressed in both policy and practice. They have signed the Student Privacy Pledge and are part of the SDPC ecosystem for legal agreements. These factors indicate a low risk of non-compliance in a Michigan deployment. (As a side note, St. Joseph Public Schools could ask Kami to sign Michigan’s own model DPA or the Common Framework through SDPC – but given Kami’s track record, they likely will sign without issue if not done already.)
5. AI Safety & Oversight
Human-in-the-Loop: Kami’s AI features are designed with teacher oversight as a fundamental requirement. Unlike an AI tutor that might interact directly with students, Kami’s content-generation AI is teacher-facing. For instance, a teacher uses Questions AI to create an auto-graded quiz; the teacher can then preview and edit every AI-generated question before assigning it. The AI never directly grades a student with finality – teachers see the autograding results and can adjust scores or feedback as needed. Similarly, if a student uses the Summarize tool on a reading passage, it’s within an assignment context where the teacher provided the material and can review the summary if desired. In short, teachers remain in control of all AI outputs that affect instruction or assessment. Kami explicitly encourages teachers to customize AI outputs (“edit them to suit your teaching style or students’ needs”) rather than blindly accept them.
Content Moderation and Bias Mitigation: The narrow scope of Kami’s AI tools inherently limits many risks. Because the AI is not answering random questions but working off provided educational content, the chance of a wildly inappropriate “hallucination” is low (it’s usually summarizing or questioning the given text). Nonetheless, there is always some risk that an AI could mis-summarize or inject a bias. Kami likely relies on the underlying AI provider’s safeguards to prevent toxic or biased outputs. Providers like OpenAI have moderation layers to avoid hate speech, sexual content, or other inappropriate material from being generated. If a teacher tried to use Questions AI on unsuitable content (say an article with extremist text), the AI would presumably refuse or sanitize the questions. Kami’s Acceptable Use Policy also forbids using the platform to generate harmful content, giving the company grounds to intervene if abuse is detected. That said, currently the teacher is the main filter – by choosing appropriate source material and reviewing AI results, they can catch any biased or nonsensical output before it reaches students.
Hallucinations & Accuracy: Recognizing that AI can sometimes produce incorrect information (hallucinations), Kami keeps AI usage focused on context-bound tasks. A summary is tied to the original document (the teacher or student can check it against the source text). Auto-generated quiz questions are derived from the curriculum provided, and teachers can double-check correctness. There isn’t an open Q\&A where the AI might fabricate an answer from thin air – it always uses the provided text or teacher’s prompt. This context grounding significantly reduces hallucination rates. Moreover, because AI use is opt-in, a cautious teacher can pilot these features and if they find inaccuracies, they can refrain or report issues to Kami. Kami’s team is likely continuously refining the AI tools based on educator feedback to improve accuracy and appropriateness. In their communications, Kami stresses that AI is a tool to enhance, not replace, teacher judgment. Teachers are encouraged to treat AI suggestions with the same scrutiny as any resource.
Bias and Fairness: Since Kami’s AI can adjust reading levels and languages, it’s being used to reduce bias and inequity (e.g., helping ELL students by translating or simplifying text without changing content meaning). The Moreno Valley study even noted how Kami’s usage helped ELL students improve significantly in reading – indicating these tools can be a force for inclusion. Of course, any AI model carries risk of subtle biases (like phrasing things in a way that favors certain cultural contexts). Teachers, by reviewing outputs, act as a bias check. For example, if the AI’s summary leaves out a perspective or oversimplifies something in a problematic way, a teacher can amend it. At present, there’s no mention that Kami’s AI has a formal bias auditing process, but using well-known AI APIs and keeping a human in loop are current best practices to manage this.
Inappropriate Output Safeguards: As mentioned, direct inappropriate output to students is unlikely given the architecture. If a student tries to get the AI to do something off-task (which they really can’t, due to no free-form chat), they won’t get far. A teacher using the AI wouldn’t prompt it with something obscene in the first place (and if they did, that’d violate terms). So the main concern would be something like the AI inadvertently including a mildly inappropriate phrase or incorrect fact in a summary or question. For this, teacher preview is the fail-safe. Additionally, Kami likely has internal testing of the AI models on educational content to fine-tune them for age appropriateness. The upcoming “Kami Companion” (if it is a conversational AI assistant) will presumably have robust content filtering and an option for teachers to see transcripts or disable it for students, to maintain safety – though details on that are forthcoming.
Teacher Overrides & Editability: A core design choice is that AI suggestions are never irreversible. Teachers can always override. Even in autograding, if the AI marks an answer wrong but the teacher sees it as acceptable, the teacher can give credit manually. The teacher can also disable AI features entirely if they sense any issue mid-lesson. For instance, if a summary tool is giving unsatisfactory results, the teacher can instruct students not to rely on it, or an admin can toggle it off for students with one click in the admin panel. This immediate control is important for safety – it’s easy to shut down an AI feature if it malfunctions or causes concern.
Transparency to Educators: Kami is transparent that AI is being used. When a teacher enables Kami’s AI, they are informed about what the AI does and that it’s optional. Any limitations (like “Questions AI works best with saved annotations”) are documented so teachers know what to expect. They’re essentially treating AI as an assistant, with the teacher as the supervisor. There is currently no fully automated decision-making that bypasses humans in Kami – everything AI does can be verified or adjusted by an educator.
In summary, Kami’s approach to AI is conservative and educator-centric. The safety net is that a knowledgeable human (the teacher) is always reviewing AI outputs before they influence students’ grades or learning materials. By limiting AI to specific tasks and giving teachers edit rights and on/off switches, Kami mitigates most risks of hallucination, bias, or inappropriate content. No AI is foolproof, but the combination of provider-level content moderation and teacher oversight means the chance of harmful AI output reaching a K–12 student via Kami is extremely low.
6. District-Level Implementations & Reviews
Kami has been widely adopted in K–12 districts across the U.S. (and globally) – as of 2025, over 40 million teachers and students use Kami in 180+ countries, with the largest user base in North America. This broad adoption suggests that many districts have vetted Kami and found it meeting their needs. Specifically, numerous U.S. school districts have rolled out Kami and reported positive outcomes:
-
Moreno Valley Unified School District (CA): In the 2022–23 school year, MVUSD did an independent study on Kami’s impact on literacy. The findings showed a significant improvement in reading scores. Students who used Kami daily saw about a 4 percentile point gain on the STAR Reading test, versus those who used it less. The effect was particularly strong for English Language Learners (ELL students benefited even more). This case study indicates Kami isn’t just convenient, but can correlate with real academic gains. MVUSD’s experience was so positive that they held professional development sessions to share best practices and are expanding use of Kami’s newer features to support literacy and formative assessment.
-
Lompoc Unified School District (CA): Lompoc USD was featured in a Kami case study (via a video) showing how they integrated Kami district-wide. Teachers there praised Kami for making digital worksheets and collaboration easy, especially during remote learning periods. They noted high student engagement and that even younger students could use the tools with ease. The district’s tech leaders appreciated Kami’s quick onboarding and integration with Google Classroom (source: Kami case study video, 2021).
-
Other District Feedback: On forums like K12 sysadmin discussions and EdTech product review sites, IT administrators often cite Kami as a “teacher-driven adoption” – meaning teachers request it for its functionality, and IT then evaluates the privacy. The consensus has been that Kami passes privacy checks (given FERPA/COPPA compliance and available DPAs) and that support from the vendor is responsive. For example, some district tech leads on Reddit mentioned that Kami filled a need for PDF editing/annotation during the shift to remote learning, and that its privacy stance was solid (no red flags in the terms) – so they proceeded to allow it. One point of caution raised was cost (free vs. paid tiers), but in terms of data privacy, districts did not report issues.
-
No Notable Public Rejections: There have been no high-profile cases of districts banning or retracting Kami for privacy or safety reasons that we could find. This is in contrast to some AI-heavy products (like certain proctoring tools or chatbots) that sparked parental complaints. Kami’s long presence (founded in 2013, in classrooms since mid–2010s) and its clean privacy record suggest districts trust it. It also helps that Kami’s functionality is largely akin to a digital notebook or worksheet tool – it’s not invasive. Even with new AI features, because they’re optional and under teacher control, we haven’t seen pushback. Districts likely pilot the AI tools and can simply opt-out if uncomfortable, rather than needing to drop Kami entirely.
-
Use Cases and Teacher Adoption: Teachers across subjects (math, ELA, science, special ed) have found creative uses for Kami. For instance, special education teachers laud Kami for its text-to-speech and annotation tools that help dyslexic students. The addition of AI “Understand” tools in 2025 was celebrated by educators of ELL and struggling readers: “The new AI functions… are excellent inclusive features especially for our non-native speaking students”. Such testimonials (like those on EdTech Impact reviews) reinforce that Kami is improving accessibility and differentiation, which is a big plus at the district level for equity goals.
-
Professional Reviews: Common Sense Media’s privacy program gave Kami a full green light, with an overall 90%+ privacy score and even an updated “Privacy Program Badge”. EdTech Impact (a UK-based review site) shows a high rating for Kami, citing improved teaching efficiency and student attainment. There are also awards – Kami has been recognized as an innovative edtech company (e.g., it was named one of the most influential companies by Time magazine in edtech, and got EdTech Digest awards). While awards themselves aren’t implementation feedback, they indicate industry confidence in the tool.
-
Technical Integration: District tech departments report that Kami integrates smoothly with Google Workspace and Microsoft 365 ecosystems – a big factor for adoption. In Google domains, deploying the Kami Chrome extension and linking it with Google Classroom is straightforward, and many Michigan schools (which often use Chromebooks) appreciate that. Some districts initially got Kami for free during the pandemic (Kami offered free licenses to schools during COVID), then later decided to purchase licenses due to teacher demand. This organic adoption story appears in many districts’ tech committee notes.
Issues or Red Flags: The absence of public controversy is notable. A few districts have mentioned cost management (ensuring they budget for the ongoing license if usage is high) but that’s not a safety issue. No known student data incidents have been reported with Kami. On the AI front, because it’s new, some district leaders are cautiously examining it – making sure it aligns with district AI policies. But given Kami’s opt-in approach, a district can simply disable AI features until their policy catches up. For instance, if St. Joseph Public Schools has not yet developed an AI use policy, they could adopt Kami but turn off the AI tools initially, and then enable them later with teacher training. Kami’s flexibility in this regard has been seen as a positive by other districts navigating the AI landscape.
In summary, district implementation experiences have been largely positive. Schools credit Kami with improving workflow (less printing, easier grading), increasing student engagement (interactive features), and now potentially boosting learning outcomes (as evidenced in literacy gains at Moreno Valley USD). The platform’s reputation among educators is strong. Crucially, there’s no pattern of privacy complaints from parents or administrators – Kami appears to have earned trust by proactively addressing privacy and by providing ample controls to districts.
7. Transparency & Governance
Policies in Plain Language: Kami’s privacy policy and terms of service are written in accessible language, relatively free of jargon. They clearly outline what data is collected and not collected, and make bold commitments (e.g., “We do not sell or share your data for advertising”). The Kami General Data Privacy Policy even prefaces with a compliance statement for COPPA/FERPA to reassure readers. For parents, Kami provides a Parent’s Guide that summarizes key privacy points in plain English (no ads, no trackers, data used only for education, etc.). This guide is a good example of transparency – it distills the legal policies into bullet points that families can understand easily. Having these multi-layered disclosures (detailed policy plus plain-language summary) is a sign of good governance.
Change Notification: If Kami makes material changes to its privacy policy or terms, they indicate they will notify users (commonly via email or in-app notice). In practice, districts using Kami typically get notified of major updates or new features through Kami’s account reps or newsletters. Additionally, because Kami signed the 2020 Student Privacy Pledge, they were committing to give prominent notice of any policy changes that affect privacy. We saw that Common Sense’s evaluation noted Kami’s policy does have a version date (so changes are tracked), though providing a change log or prior versions wasn’t assessed. As of now, there haven’t been any sudden policy shifts – the company’s stance on privacy has remained consistently strict (if anything, moving toward even more transparency over time).
Transparency Reports: While Kami doesn’t publish a “transparency report” in the way big tech companies do (e.g., reporting government data requests or content moderation stats), it likely hasn’t needed to – it’s not a platform for user-generated public content, and government requests for education data are rare. Kami does promise in contracts to notify if government entities request student info, giving the district a chance to object (this is often in DPAs). There’s no indication of any such requests happening publicly.
Product Updates & Roadmap Disclosure: Kami is quite open about its product roadmap and updates, which is a form of transparency to users. They maintain a “Kami Updates” blog/channel announcing new features (e.g., the introduction of the Summarize tool and other AI features in April 2025 was publicized). They also held a Forward Focus webinar in May 2025 where the CEO discussed the vision and upcoming features. This lets schools know what’s coming (such as the Kami Companion AI), so there are no surprises. When they roll out AI beta features, they typically label them as beta and optional. This approach ensures that schools are aware and can prepare (for example, inform teachers or adjust settings) rather than discovering new AI functionality accidentally.
Governance and External Oversight: Kami subjects itself to third-party reviews, which improves trust. For example, undergoing the Common Sense Privacy Assessment (and earning their seal) means an external body examined Kami’s policies for any gaps. Kami also engages in the Student Data Privacy Consortium (SDPC), signing model contracts that are publicly posted on the SDPC site. Many of those agreements (which we have referenced) are publicly accessible, effectively allowing anyone to inspect the privacy promises Kami made to various districts. This level of openness – having legally binding documents in the public domain – holds Kami accountable to what it proclaims. Additionally, Kami’s participation in conferences and forums (their team frequently speaks about student data privacy and AI ethics in edtech) suggests they contribute to the community dialog on governance.
User Controls and Transparency to End-Users: Within the app, Kami provides some transparency features: teachers can see a log of annotations to know who did what and when (the “Annotation Summary” acts like a history log) – so there’s accountability in collaborative use. There’s also an insights dashboard for teachers giving engagement statistics (e.g., how many annotations each student made), which helps teachers identify if a student might not be engaging. These aren’t “transparency reports” in a privacy sense, but they do make the system’s activity transparent to educators, which can indirectly surface any misuse.
Handling of AI Rollout: A key governance question is how Kami handles introducing AI updates. So far, they’ve done this cautiously: new AI features have been rolled out as opt-in betas with admin controls. They have also provided documentation and even myth-busting blog posts to educate stakeholders. For instance, Kami’s blog addressed concerns in “Myth buster: AI isn’t safe for schools”, explaining how they mitigate risks and promising continued focus on student safety. This kind of communication is essential for governance in the era of AI – being upfront about what the AI does, doesn’t do, and how they keep it in check.
Compliance Documentation: Kami’s governance includes keeping necessary documentation up to date, like Data Processing Agreements (for GDPR) and state-specific exhibits. They appear willing to undergo audits: Louisiana’s contract explicitly allows the state to audit Kami’s data security practices, and Kami agreed to that. This indicates that, if MI law or the district requires a security questionnaire or audit, Kami would cooperate fully.
In summary, Kami demonstrates strong transparency and governance practices for a K–12 vendor. They communicate clearly with schools about privacy and feature changes, provide accessible policy summaries for laypersons, and engage in external privacy initiatives for accountability. This culture of transparency reduces the likelihood of any unpleasant surprises for St. Joseph Public Schools – Kami will keep the district informed and involved in how the platform is used and evolves.
8. Evaluation Matrix
Below is an evaluation matrix scoring Kami on each dimension (1 = poor, 5 = excellent). Each category is scored 1–5, with notes on strengths and any red flags:
| Category | Score | Notes & Key Findings |
|---|---|---|
| AI Functionality & Purpose | 🟢 5 | Robust, teacher-focused AI toolkit (auto-grading, summarizing, translating, etc.). Features clearly target educational outcomes (saving time, personalizing learning). Optional opt-in use prevents unwanted AI interactions. Kami’s AI is an enhancement layer, not a replacement for teaching, which aligns well with K–12 needs. |
| Data Handling & Privacy | 🟢 5 | Exemplary data practices. Collects minimal student info; no student files stored on Kami servers (annotations sync via Google/Microsoft cloud). Uses secure AWS/GCP infrastructure in US. No ads or sale of data. Strong encryption in transit & at rest. Offers deletion (right to be forgotten) on request. No red flags – meets or exceeds industry standards. |
| Safeguards by Age Group | 🟢 4 | Good role-based controls for features (can disable AI or other tools for students vs. teachers). COPPA compliance via school consent. Interface and tools are age-appropriate; no open social features that pose risk. Real-time teacher monitoring (Class View) adds safety for all ages. Could improve by providing more automated alerts for concerning content (currently relies on teacher observation). Overall, appropriate protections are in place for elementary through HS. |
| Regulatory Compliance | 🟢 5 | Fully compliant with FERPA, COPPA, SOPIPA (and by extension Michigan’s PA 367 privacy requirements). Privacy policy explicitly aligns with these laws. Has signed the national Student Privacy Pledge. Will sign DPAs; example NDPA contracts show adherence to all required clauses (data ownership, breach notice, etc.). Common Sense Privacy seal achieved. No legal compliance issues identified. |
| AI Safety & Oversight | 🟢 4.5 | Strong human-in-the-loop approach – teachers vet all AI outputs. AI features are constrained to specific tasks (reducing risk of misuse or harm). Admins can turn AI off if needed. Likely uses vetted AI models with content filtering, though specific model transparency could be better (unclear which AI engine is used). No direct student-to-AI chat, which is safer for now. Minor risk of AI inaccuracies remains, but mitigated by teacher review. Overall, Kami has implemented AI cautiously and responsibly. |
| District Implementations | 🟢 5 | Widely adopted with positive results. Case studies (e.g. Moreno Valley USD) show academic improvement with Kami use. No known privacy incidents or community pushback – indicates districts trust Kami. Integration is smooth with Google/Microsoft environments (important for Michigan schools). High teacher satisfaction and usage suggests Kami delivers on its promises. Strength in numbers: 40M+ users and growing, yet no negative patterns reported. |
| Transparency & Governance | 🟢 5 | High transparency. Privacy terms are clear and publicly available, plus parent-friendly summaries. Regular updates on new features, with opt-in choices and documentation. Participates in external privacy evaluations and pledges. Breach notification practices are in place. Kami is proactive in addressing AI concerns in blogs and communications. Governance appears strong, with continuous engagement with privacy standards bodies (e.g., SDPC, Common Sense). No governance red flags. |
Key: Green scores (4–5) indicate strengths; Yellow (3) would indicate caution; Red (1–2) would indicate serious concern. Kami scored in the green across all categories in our evaluation, reflecting a well-rounded and trustworthy platform for classroom use.
9. Final Recommendation
Strengths: Kami demonstrates a comprehensive commitment to student safety, privacy, and educational value. Its core functionality (digital annotation, collaboration, feedback) is enriched by well-thought-out AI tools that save teachers time and help differentiate learning. Privacy practices are top-tier – no advertising, no selling data, and storage of student work remains on district-controlled cloud drives. Compliance with all relevant laws (FERPA, COPPA, state laws) is documented and vetted. Teachers remain in control of AI outputs, and admins have fine-grained control over feature access, allowing the district to introduce AI features at a comfortable pace. Kami’s large user base and positive case studies (including improved reading outcomes) show it can be deployed at scale with real benefits. The company has a proven track record since 2016 with no known security or privacy incidents, and it engages in transparent communications and agreements.
Concerns: The concerns are relatively minor. AI features are new, and while optional, they require thoughtful rollout – teachers will need training to use Questions AI or Summarize effectively and ethically. The district should ensure that any use of AI complies with local policies (e.g., if a policy requires notifying parents about AI, the opt-in nature helps address that). Another consideration is continued monitoring of third-party AI providers – Kami doesn’t explicitly name them, so St. Joseph PS may want to ask Kami for assurances (in writing) that any AI service used will not store or misuse district data (Kami likely has this covered, but it’s good due diligence). Additionally, device integration: students will primarily use Kami via Chromebooks; the district should confirm the Chrome extension and Drive integration are set up securely (Kami provides guides, and it’s straightforward, but IT should manage the extension deployment to prevent rogue look-alike extensions). Finally, cost is not a privacy issue but a practical one – if Kami’s full features require licensing, the district should plan for sustainability of funding (the value is high, so this is usually justified by usage).
None of these concerns are show-stoppers; they are typical implementation checkpoints. There are no red flags in Kami’s privacy, legal compliance, or safety profile. The few cautionary notes (new AI tools, etc.) can be managed with the controls and knowledge we have.
Recommendation: 🟢 Go for adoption. Kami is green-lit for K–12 deployment in St. Joseph Public Schools. The platform’s strengths in enhancing digital learning and its rigorous privacy safeguards far outweigh any minimal risks. It earns a “Go” recommendation – with a suggestion to proceed with a phased rollout of the AI features (perhaps start with teacher use and pilot classrooms) and clear communication to parents about the tool (leveraging Kami’s parent guide to address any privacy questions). Overall, Kami appears to be a valuable and safe addition to the district’s edtech toolkit, aligning well with Michigan’s student privacy requirements and the district’s instructional goals.
10. Key References & Resources
- Kami Privacy Policy: Kami Student Data Privacy Policy (COPPA & FERPA) – Official policy outlining data practices, FERPA/COPPA compliance, and privacy commitments.
- Kami Terms of Service: Kami Terms of Service – User agreement detailing service usage terms, including age requirements and data rights.
- Student Privacy Pledge Certification: Kami Limited was a signatory of the Student Privacy Pledge (legacy signatories list) – demonstrating public commitment to K–12 privacy principles.
- Sample Data Privacy Agreement: Standard NDPA with Kami (Notable Inc.) – Feb 2023 – Example of a signed Student Data Privacy Consortium agreement between Kami and a U.S. school, showing contract terms (encryption, data deletion, etc.).
- Common Sense Privacy Evaluation: Common Sense Media – Kami Privacy Report – Independent evaluation of Kami’s privacy policies (Kami earned the Privacy Verified seal).
- Case Study – Reading Outcomes: Empirical Education Study on Kami (Moreno Valley USD, 2025) – Research report showing a 4 percentile point improvement in reading scores with daily Kami use (useful to share with stakeholders interested in AI’s impact).
- Kami Parent/Guardian Guide: Kami Help Center – Parent’s Guide – Explains in non-technical terms how Kami works and protects data, which can be shared with families for transparency.
- Kami & AI Announcement: Kami Updates – April 2025 – Blog post outlining the new AI features (Summarize, etc.) and the upcoming “Kami Companion,” useful for understanding how Kami introduces AI advancements. (Note: If inaccessible, similar info is available via Kami’s YouTube update for April 2025.)
- SDPC Listing (MI example): St. Joseph PS can reference the Michigan NDPA. While not provided here as a link, the district can use the Michigan Student Privacy Alliance portal to execute Kami’s statewide agreement (if available) or adapt an existing one like the above NDPA. Kami is listed under “Notable Inc. (dba Kami)” in many state privacy registries.
Comments
0 comments
Please sign in to leave a comment.