Scope and Status Note
This report covers the IBM SkillsBuild Artificial Intelligence Certificate only. The Cybersecurity and Data Analytics certificates are referenced solely where their precedent bears on the AI certificate. Nothing here is a proposal to adopt those.
This is an independent analysis prepared by community college faculty. It is not authored, reviewed, or endorsed by IBM, and it does not speak for IBM. Where a fact could not be verified from a primary source it is marked as unverified rather than asserted.
Reader's Note
This report is written to help the certificate get adopted, not to argue against it. The content is well built and, where we have tested it directly, it holds up. What follows is mostly about the layer around the courseware — the wrapper a California college must add, and a short list of questions only IBM can answer. Read the length of that list as a measure of California's regulatory detail, not of the product's quality.
Executive Summary
The IBM SkillsBuild Artificial Intelligence Certificate is approximately 72 hours of self-paced, asynchronous online learning: 13 modules and 4 embedded capstone projects grouped into 4 microcredentials, plus a culminating certificate, delivered on skills.yourlearning.ibm.com and credentialed through Credly. It is free to learners, requires no prerequisites beyond basic computer literacy, and is aimed at non-technical adults — explicitly including faculty.
Three things determine whether it can be adopted, and none of them is about content quality.
First, there is no ACE credit recommendation for this certificate. ACE's National Guide lists exactly two IBM SkillsBuild entries — Cybersecurity (IBM-0028, 12 semester credits) and Data Analytics (IBM-0029, 8 credits), both expiring 30 June 2027. There is no Artificial Intelligence row. Because ACE's registry is precisely where such a recommendation would appear, its absence is dispositive rather than merely silence. IBM's own faculty playbook states that credit recommendations for the AI certificate are "in progress," but attaches no date. The practical consequence: Title 5 §55050(d), which obliges colleges to consider ACE recommendations, is an empty hook for this product today.
Second, and decisively, the certificate as IBM delivers it cannot by itself satisfy regular and substantive interaction. Federal law (34 CFR §600.2) and California law (Title 5 §55204) both require that a distance course include substantive interaction between the instructor and students comprising at least two of five enumerated activities. Every one of those activities is framed as conduct by the instructor. IBM's pre-recorded modules and auto-scored assessments are not instructor conduct, and therefore supply none of them. A course whose only interaction is the vendor platform fails RSI and risks reclassification as correspondence education, which by rule is not distance education at all.
Third, IBM has published no accessibility conformance report for this platform. IBM operates a mature public ACR portal covering roughly 4,100 offerings. Filtering it for "SkillsBuild" returns zero results; "watsonx" returns 152. The college's own signoff standard requires a VPAT, an internal accessibility review, or an approved accessible alternative before a third-party platform can be made mandatory. That evidence does not currently exist and must be produced.
None of these is fatal. All three are addressable. But they determine the order of operations, and they mean the fastest route to students is not the same as the most durable one.
Recommendation. Run two tracks in parallel. Near-term, embed the certificate as courseware inside an existing approved course with a faculty wrapper designed explicitly to satisfy RSI, and file a Credit for Prior Learning mapping in the same senate cycle. Durably, build a noncredit CDCP Short-term Vocational sequence from the four microcredentials on a 12–18 month timeline. Both routes require the accessibility evaluation first, because both make a third-party authenticated platform a required element of instruction.
What the Certificate Actually Is
| Total learner effort | ~72 hours (IBM); 73.7 hours summing every unit in IBM's syllabus |
| Structure | 13 modules + 4 capstone projects, in 4 microcredentials |
| Credentials | 5 Credly badges (4 microcredentials + culminating certificate) |
| Delivery | Asynchronous, self-paced, online |
| Prerequisites | Basic computer literacy; none subject-matter |
| Audience | Non-technical adult learners, students, and faculty |
| Platform | skills.yourlearning.ibm.com (plus IBM watsonx Prompt Lab for the final capstone) |
| Access | Invitation-gated; requires institutional registration and IBM adding the college to a private group space |
| Scored work | 6 items only — 4 capstones, 1 final assessment, 1 final capstone |
| Attempts | Unlimited, 80% to pass |
| Modules | Completion-based, not graded |
The four microcredentials each carry one learning outcome and one capstone, which makes them unusually well-suited to serving as a course sequence. That structural fact does most of the work in the noncredit route below.
A precision point that matters more than it looks. IBM's 72 hours is total learner effort, in IBM's own words — not contact hours. Under the standard lecture ratio, a credit course claiming 72 total hours is roughly 24 contact hours plus 48 outside-of-class. At the Chancellor's Office recommended divisor of 54 total student-work hours per unit, 72 hours yields 1.33 units; at the 48-hour minimum divisor, 1.5. Presenting 72 as contact hours and deriving four or more units would inflate the calculation roughly threefold and is the single most likely technical error to sink a curriculum submission. The divisor must be taken from adopted district policy, never assumed.
What Is Actually Being Adopted: The Delivery Chain
Walking one module in full — all 21 of its lesson views — established something none of the documentation discloses. Delivering this certificate involves six hostnames across three organisations.
| Host | Organisation | Role |
|---|---|---|
skills.yourlearning.ibm.com |
IBM | Catalogue, learning plan, progress. The only host named anywhere in IBM's materials. |
alm.ibm.com |
IBM | IBM's Adobe Learning Manager tenant. Launches modules. |
learningmanagerfr.adobe.com |
Adobe | Adobe Learning Manager, EU regional endpoint. |
cpcontents.adobe.com |
Adobe | Adobe Captivate prime-player, build tagged prodfr. Hosts the authored course. |
video.ibm.com |
IBM | Embedded video within lesson views. |
csr.h5p.com |
H5P | Embedded interactive activities. A third company entirely. |
The actual learning platform is Adobe Learning Manager, not IBM, the regional endpoint is European, and the interactive activities come from H5P. None of this appears in public documentation, in IBM's faculty enablement playbook, or in the certificate materials.
It changes three things at once.
The accessibility evaluation has to cover the whole chain. A conformance report for any single layer would not describe what a student experiences. Each layer contributes its own controls, focus behaviour, and markup.
The FERPA and data question gets bigger and more specific. Whatever the institutional agreement says about IBM, it now has to answer for Adobe and H5P as sub-processors and for data residency in an EU region. That moves from a generic "review the vendor terms" item to a named question that belongs in the list for IBM.
The remediation question is settled, and the answer is no. The content layer is cross-origin from IBM's wrapper. The institution cannot inspect, instrument, or fix anything in it. Its only remedies are to ask IBM or to provide an alternative path. The research raised this possibility in the abstract; it is now demonstrated for this product.
Findings from the one module examined
Every embedded media and interactive frame is unnamed. Sweeping 21 lesson views of module 1 found four embedded frames — one video, three interactive activities — and not one carries a title attribute. A 100% failure rate.
These are not decorative frames. A screen reader user navigating by frame reaches four unnamed regions and is told nothing about what any of them contains. This maps to WCAG 4.1.2 Name, Role, Value (technique H64), and it is the most consequential accessibility finding so far because it affects the interactive and media content directly.
The wrapper naming defect is systematic. Module 1 is framed as title="course:4068807", module 2 as title="course:4068808". Two for two — this is how IBM's integration layer names every module, not one bad frame. Adobe's own layers name theirs correctly, including marking decorative frames Intentionally blank, so the defect sits squarely in the layer IBM controls.
The authored content is genuinely well built, and this report says so plainly. Across the 21 lesson views sampled: 67 images, none missing alt text; exactly one <h1>; clean heading order; lang set; 49 focusable elements. The START MODULE control renders as an <a> rather than a <button> — a frequent keyboard trap in authored courseware — but carries a real href and is properly focusable. A second module matches on every structural measure.
The pattern is careful authoring and defective integration. That matters for where remediation pressure belongs: on IBM's wrapper and embed configuration, not on the instructional designers.
A permissions observation for the security review. Every H5P activity frame carries allow="autoplay *; geolocation *; microphone *", delegating those permissions to a third-party origin. Whether H5P exercises any of them is not established and it may be an embed-template default — but it is a concrete question for IBM.
One result deliberately not treated as a finding
The browser accessibility tree read from IBM's wrapper returned essentially nothing — a single generic node with no content beneath it. Read carelessly, that says the course is invisible to assistive technology.
It almost certainly means the inspection tool did not traverse a cross-origin frame. It is recorded here specifically so that nobody later reports it as a defect. Settling it requires a real screen reader, which is the only authoritative check.
An observation for the security review, not this one
The Captivate player is invoked with an access_token in the URL query string. The value is deliberately not recorded anywhere in this analysis.
Bearer tokens in URLs are exposed to browser history, referrer headers, and server logs. The token may be short-lived, single-use, or narrowly scoped — none of which is visible from the client. This belongs to whoever performs the district's security review. It is not a vulnerability claim.
Scope of this section
Observed 27 July 2026 on a learner account, viewing without completing. Twenty-one lesson views of module 1 (MC1, AI Impact) were sampled, and module 2 sampled and found consistent. That is an incomplete subset of module 1, not the whole module — the Rise player unmounts off-screen sections, so a DOM count at one moment undercounts; the module contains further lessons that were not examined. The capstones, the final assessment, watsonx Prompt Lab, and the Credly claim flow remain unexamined, as do the keyboard-only, screen-reader, 400% reflow, contrast, and text-spacing passes. Nothing in this section is a conformance determination.
Sampling note: WCAG-EM 2.0 asks for a representative sample of views and states, not every page. Modules 1 and 2 match on every structural measure, which suggests a single consistent template — so walking the remaining eleven module landing views would add little. The unexamined risk sits in the distinct view types listed above, and that is where the evaluation should go next.
The Five Routes, Ranked
Ranked by speed to students against approval burden. Each names the body whose approval it requires.
1. Embed as courseware in an existing approved course. Department and discipline faculty only; no new curriculum action if the course outline's outcomes and hours are unchanged. Term-scale — the fastest credit-bearing route. Costs: no new credential recognition, and it triggers the no-cost-to-student and accessibility reviews because it makes a third-party authenticated platform required. Constrained by RSI: Title 5 §55204(a) applies to "any portion" of a course conducted at a distance, with no de minimis carve-out, so the faculty wrapper is mandatory even when only the IBM modules are asynchronous.
2. Not-for-credit community education. Local administration only; fastest overall. Costs: no credit, no apportionment, and typically a student fee — which cuts directly against the equity argument that otherwise makes this certificate attractive. The approval mechanics for this route were not verified against primary sources and should be confirmed against district policy before being relied on.
3. Credit for Prior Learning under §55050. Requires the board's existing CPL policy (already mandated), curriculum-committee-approved procedures, and discipline faculty designing the assessment. Awards credit for the credential without teaching it — which sidesteps the RSI problem entirely, since no distance course is being offered. Blocked from the ACE shortcut, so it must run as a local challenge exam or portfolio against an existing catalog course outline. §55050(d) permits credit only for "individually identified courses" already listed in the catalog, so a host course outline must exist first. The four capstones plus the final capstone are the natural portfolio evidence.
4. New noncredit CDCP Short-term Vocational sequence. Curriculum committee, governing board, and Chancellor's Office program approval, plus an EDD-consulted "high employment potential" determination. Slowest, and the only route that is simultaneously free to students and revenue-positive to the district at the credit apportionment rate. CDCP requires a sequence of two or more courses leading to a Certificate of Completion — the four microcredentials map onto that requirement cleanly, which is the strongest structural argument in this report. Note that of CDCP's four eligible domains only Short-term Vocational is available: "Workforce Preparation" is defined as pre-vocational basic skills, i.e. preparation to enter technical training, not technical content itself.
5. New standalone credit Certificate of Achievement. Curriculum committee, board, and Chancellor's Office. Weakest fit — 1.0–1.5 units is far below a viable standalone certificate. Only sensible as one course inside a larger locally built program.
The Interaction Problem
This is the section that determines whether routes 1 and 4 are viable, and it deserves to be read closely.
What the law requires
34 CFR §600.2 defines substantive interaction as "engaging students in teaching, learning, and assessment, consistent with the content under discussion," and requires at least two of the following: providing direct instruction; assessing or providing feedback on a student's coursework; providing information or responding to questions about the content; facilitating a group discussion regarding the content; or other instructional activities approved by the accreditor.
"Regular" interaction has two requirements joined by and: providing the opportunity for substantive interaction "on a predictable and scheduled basis commensurate with the length of time and the amount of content," and monitoring the student's academic engagement and success with prompt, proactive outreach. When commenters asked the Department to change that "and" to "or," it declined explicitly.
California imports this almost verbatim at Title 5 §55204, amended effective 4 December 2022. Two features of the California text matter especially:
- §55204(a) applies to "any portion" of a course conducted through distance education. There is no minimum threshold.
- §55204(d) declares that "Regular and substantive interaction is an academic and professional matter pursuant to sections 53200 et seq." This is the strongest citation available for Senate primacy over this decision — it makes the Senate's authority here regulatory rather than customary.
Why the vendor platform cannot supply it
All five substantive-interaction elements describe acts of the instructor. IBM is not the instructor of record. A vendor auto-scored knowledge check is not the instructor "assessing or providing feedback on a student's coursework." IBM's own faculty playbook confirms the capstones and final assessment are auto-scored. The platform therefore contributes zero of the required two elements.
IBM's playbook is candid about its model: "Faculty support learner progress rather than deliver traditional instruction," with faculty "acting as facilitators." Mapping IBM's five stated faculty contributions against the five required elements:
| IBM's stated faculty contribution | Satisfies a §55204(b) element? |
|---|---|
| Supporting learner progress — encourage participation, reinforce momentum | No |
| Clarifying concepts — address questions, resolve confusion | Yes — element (3) |
| Guiding application — encourage structured thinking | Arguable at best |
| Encouraging engagement — check-ins, reminders | No |
| Connecting learners to support systems | No |
IBM's model supplies at most one of the two required elements, and that one is reactive rather than scheduled. This is not a criticism of IBM's design; the certificate is built for a global, largely non-credit-bearing audience where RSI does not apply. It is a statement of what a California credit deployment must add.
What failing RSI costs
A distance course that fails RSI does not become unclassified. It matches the §600.2 definition of a correspondence course — "instructional materials, by mail or electronic transmission, including examinations on the materials, to students who are separated from the instructors," where interaction "is limited, is not regular and substantive, and is primarily initiated by the student" — and the rule states plainly that "a correspondence course is not distance education." The SkillsBuild fact pattern maps onto that definition uncomfortably well.
This should not be overstated, and the report will not overstate it. The Title IV consequences at 34 CFR §600.7 are institution-level thresholds measured across an award year: more than 50% of courses, or 50% or more of regular enrolled students. One RSI-failing course comes nowhere near that. The honest framing is that RSI failure is an audit and program-review exposure at the course and apportionment level, and that a systemic pattern of vendor-wrapped RSI-failing courses is what implicates institutional eligibility.
IBM's three delivery models against California law
IBM's playbook offers three models. Only one survives.
| IBM's model | IBM's description | Assessment |
|---|---|---|
| Independent | "Fully self-paced with minimal faculty involvement" | Fails RSI. Not viable for credit. |
| Hybrid | "Self-paced learning with periodic check-ins or facilitated sessions" | Marginal. "Periodic check-ins" is not evidently a "predictable and scheduled basis commensurate with the length and amount of content." Viable only with a published, enforced cadence. |
| Fully embedded | "Integrated into a course with ongoing faculty support" | The only defensible model, and only if the support is specified as scheduled, documented acts. |
What the faculty wrapper must do
To reach two elements, the defensible pairing is faculty-scored capstones — a faculty-authored rubric, faculty-graded, not vendor-scored, satisfying element (2) — combined with either faculty-facilitated discussion of module content on a published cadence (element 4) or faculty-delivered direct instruction supplementing the modules (element 1). The cadence must be published in the syllabus and the distance education addendum.
What does not count: vendor auto-graded quizzes; vendor-generated feedback; a syllabus alone; an inactive instructor; and "available upon request" — because "upon request" appears only in §55204(c)(2) as a trigger for the responsive duty, never as a substitute for (c)(1)'s independently scheduled opportunity.
A structural obstacle created by the gating. Because the platform is invitation-gated behind IBM's login, faculty monitoring data may live entirely outside the college's LMS and outside the college's control. If the college cannot obtain a data feed, or require students to submit evidence into the college LMS, then §55204(c)(2) monitoring cannot be performed at all. This is a question for IBM before any approval, not after.
Time sensitivity. A CCCCO 45-Day Notice published 30 April 2026 proposes amending Title 5 §§55200, 55204, and 55210, with a hearing on 19 May 2026 and comments closing 14 June 2026. Whether the Board adopted, modified, or deferred the action could not be confirmed. Any Senate approval should therefore reference §55204 "as amended from time to time" rather than freezing a citation, and no proposed language should be quoted.
The Assessment Model
The certificate contains six scored items: four capstone projects, one final assessment, and one final capstone. All permit unlimited attempts at an 80% pass mark. The 13 modules are completion-based. Practice activities are scored but, in IBM's words, "not activities that count towards a grade."
IBM's playbook describes the design as "auto-scored capstone projects and final assessments" that "introduce structured evaluation, ensuring consistency and rigor while allowing unlimited opportunities for skills acquisition," and markets it as letting institutions adopt the certificate "without requiring significant changes to assessment or evaluation systems."
For a California credit-bearing deployment, changes to the assessment system are precisely what compliance requires. Two distinct issues follow, and a Senate will raise both:
Assessment validity. Auto-scoring combined with unlimited attempts means a student who passes on the first attempt and one who passes on the twentieth receive an identical credential, with no visible distinction. Whether the institution can even see attempt counts and per-attempt scores is unknown and must be asked.
Faculty authorship of assessment. Under §55050(c), where CPL is used, "the nature and content of the assessment shall be determined solely by faculty in the discipline who normally teach the course." Vendor-authored, vendor-scored assessment cannot discharge that duty. Under the embedding route, the same logic drives the wrapper design: if faculty want element (2) to count toward RSI, they must score the work themselves.
The recommendation in both cases is the same. Treat IBM's capstones as learning artifacts and assess them locally against a faculty-authored rubric. This resolves the validity objection and the RSI gap with one mechanism.
Accessibility Position
The college's existing signoff standard conditions adoption on required third-party platforms having "a VPAT, internal accessibility review, or approved accessible alternative," at a WCAG 2.2 AA target with 2.1 AA as the minimum baseline.
IBM publishes no ACR for this platform. IBM operates a public, unauthenticated conformance report portal at ibm.com/able/product_accessibility/ covering roughly 4,100 offerings and publishing reports in VPAT 2.x format. Filtered by hand on 27 July 2026:
watsonx→ 152 reportsSkillsBuild→ 0 resultsPrompt Lab→ 0 results
The watsonx result confirms the filter functions, so the zeroes are a genuine searched-and-not-found. Research located no report for IBM Your Learning, IBM Training, or Credly in the catalog either.
This is worth stating carefully. It is not evidence that the platform is inaccessible. IBM commits its own teams to WCAG 2.2 Level AA under the IBM Accessibility Requirements v7.3 — a higher bar than the DOJ rule's WCAG 2.1 AA. The gap is in published evidence, and the asymmetry is what makes it worth raising: watsonx, which this certificate's final capstone requires, has 152 published reports; the courseware wrapping it has none.
Credly does publish a VPAT, but it should not be relied on: VPAT 2.4, dated 26 May 2022, scoped only to WCAG 2.0 A/AA with the WCAG 2.1 rows left as unedited template placeholders, and self-reporting 19 "Does Not Support" findings. It does not attest to the standard the DOJ rule adopts.
What direct testing found
Rather than rely on documentation, the courseware itself was tested on a learner account. On every criterion tested so far, the authored content passes.
| Criterion | Result | Evidence |
|---|---|---|
| 1.1.1 Non-text Content | Passes | 67 images sampled, zero missing alt text |
| 1.2.2 Captions (Prerecorded) | Passes | Captions provided and verbatim accurate, correctly punctuated and capitalised — consistent with human authoring rather than raw speech recognition |
| 2.1.1 Keyboard | Passes | Checkbox questions operate with Space; the drag-and-drop activity is keyboard operable — Space grabs an item and the drop zones become live targets |
| 2.4.7 Focus Visible | Passes | Clear high-contrast focus ring throughout; the page scrolls to bring the focused element into view |
| 2.5.7 Dragging Movements | Untested | Requires a single-pointer alternative; keyboard support does not satisfy it. Likely present, not verified |
| 4.1.2 Name, Role, Value | Needs a fix | Embedded media and interactive frames carry no title attribute |
The result worth highlighting is the drag-and-drop. A drag activity with no keyboard path is the most common hard blocker in authored courseware, and this one has a keyboard path: Tab reaches each item, Space picks it up, and the category zones become live drop targets. The instruction wording — "Match each task… then select Submit" — avoids both "drag" and "click", which is deliberate good practice.
Two refinements would improve it: Escape does not cancel a grab, and the keyboard interaction is never explained on screen, so a keyboard user must discover it. Both are small changes.
The 4.1.2 issue and where it sits. Across the module sampled, four embedded frames — one video and three interactive activities — carried no title, and IBM's wrapper frames each module with an internal identifier (course:4068807, course:4068808). Adobe's own layers name their frames correctly, including marking decorative ones Intentionally blank. So this is embed configuration in the layer IBM controls, not an authoring problem, and it is a small fix with a real benefit to screen reader users.
Scope. This covers 21 lesson views of one module plus samples of another — an incomplete subset, since the Rise player unmounts off-screen sections and the module contains further lessons. The capstones, the final assessment, watsonx Prompt Lab, and the Credly claim flow are unexamined, as are the screen-reader, 400% reflow, and contrast passes. Nothing here is a conformance determination.
The regulatory deadline. DOJ's ADA Title II web rule adopts WCAG 2.1 Level AA. An Interim Final Rule published 20 April 2026 (91 FR 20902) extended both compliance tiers by one year: 26 April 2027 for public entities of 50,000+ other than special district governments, and 26 April 2028 for entities under 50,000 or any special district government regardless of population. Neither deadline has passed. Which tier applies to a California community college district turns on whether it is a special district government — a classification this report does not resolve and which should be settled by district counsel rather than inferred.
Because IBM has published no report, the college must either produce its own independent evaluation or approve an accessible alternative. A companion evaluation is in preparation. It will be titled an independent accessibility evaluation, not an Accessibility Conformance Report — ITI and Section508.gov both frame ACR authorship as vendor self-assessment about the vendor's own product, and a college cannot issue an ACR for software it does not own. It will use the VPAT 2.5Rev structure and WCAG 2.2 A/AA criteria, and will state plainly that it is not authored or endorsed by IBM.
Two environments require separate evidence, not one: the courseware player and IBM watsonx Prompt Lab.
What We Need From IBM
These are the questions that block a Senate recommendation. Most cannot be answered from public sources because the institutional agreement is not public — skillsbuild.org/terms-of-use and /privacy both return HTTP 404.
- ACE. When is the Learning Evaluation review of the AI certificate expected to conclude, and what recommendation is being sought? A pending submission changes which route we file first.
- Assessment transparency. Are per-learner attempt counts and per-attempt scores on the capstones and final assessment visible to the institution?
- Monitoring data. Is there a faculty or administrator role with a progress dashboard, and can engagement data be exported into the college LMS? Without it, §55204(c)(2) monitoring cannot be performed and the embedding route fails.
- LMS integration. IBM's own recruiting for a SkillsBuild-for-Academia technical lead references implementing "LTI integrations (LTI 1.3) that connect SkillsBuild offerings with partner LMS systems (Canvas, Moodle, Blackboard)" and familiarity with SCORM, xAPI, and Common Cartridge. Does that integration exist for partner institutions, and is it available here? Nothing about it is publicly documented.
- Accessibility. Is there a VPAT or ACR for
skills.yourlearning.ibm.com, watsonx Prompt Lab, or Credly — published or available under agreement? - The agreement itself. What does the institution sign, and what does it say about student data ownership, FERPA roles, reuse of IBM content inside a college LMS, and what happens to already-issued Credly badges if the institution exits?
- California precedent. IBM's July 2024 announcement names the Bay Area Community College Consortium as a deployment partner for the Cybersecurity and Data Analytics certificates. What articulation, MOU, or CPL language did those colleges execute? That is the nearest California template and obtaining it is a phone call.
Two smaller items observed in IBM's own materials, offered as constructive feedback: the AI Final Capstone's assessment details state that learners will "demonstrate your knowledge of organizational cybersecurity and incident response," which appears carried over from the Cybersecurity certificate; and the learning plan carries a Level tag of "Advanced" while the course description targets non-technical learners new to AI.
Decision Position
The certificate is worth adopting. The content is coherent, the microcredential structure is unusually well-matched to California's noncredit sequence requirements, it is free to students, and it addresses AI fluency for exactly the non-technical population that most needs it.
It cannot be adopted unchanged — but the changes are ordinary curriculum work, not obstacles. Of IBM's three delivery models only Fully embedded satisfies California's interaction rules, and auto-scored capstones cannot by themselves discharge faculty responsibility for evaluating student work. The college closes both with a faculty wrapper: a published interaction cadence, locally scored capstones, and engagement evidence inside the college LMS. None of that asks IBM to change the product.
What the college cannot supply on its own is the evidence. Three things should be in hand before a Senate votes: accessibility documentation for the delivery stack; a written answer from IBM on monitoring data and attempt visibility; and the institutional agreement, so that data, FERPA, licensing, and credential-portability terms are reviewed rather than assumed. All three are straightforward for IBM to provide, and none of them requires a product change.
Recommended sequence: commission the accessibility evaluation now, since it gates both viable routes; put the seven questions to IBM in writing; and file the embedding and CPL routes together in the next Senate cycle, with the CDCP sequence as the 12–18 month objective.
Sources
Primary sources relied on, with the caveat that regulatory citations should be pinned to official text before publication — govt.westlaw.com/calregs, casetext, and asccc.org were unfetchable or returned 403 during research, and Title 5 §55050 and §55204 were verified through Cornell LII.
- 34 CFR §600.2 (distance education; substantive interaction; correspondence course), §600.7 (institutional eligibility thresholds) — verified against the eCFR 2026-07-01 snapshot and GPO CFR XML
- 85 FR 54809 — Distance Education and Innovation final rule, 2 September 2020, effective 1 July 2021, including the preamble declining to change "and" to "or"
- Title 5 CCR §53200 (academic and professional matters), §53203 (consult collegially), §55002 and §55002.5 (course standards and units), §55050 (Credit for Prior Learning), §55151 (noncredit programs), §55204 (instructor contact, amended operative 4 December 2022)
- CCCCO Program and Course Approval Handbook, 8th Edition; CCCCO Hours and Units Calculations instructions; CCCCO 45-Day Notice of Proposed Rulemaking on Distance Education, 30 April 2026
- Education Code §§84750.4, 84750.5, 84760.5; SB 361 (Scott, 2006, Ch. 631)
- ACE National Guide — IBM organization page; IBM-0028 and IBM-0029 course detail pages
- 28 CFR §35.200; DOJ Interim Final Rule, 20 April 2026, 91 FR 20902
- ITI VPAT 2.5Rev (April 2025), WCAG edition; Section508.gov ACR guidance
- W3C WCAG 2.2 (Recommendation, updated 12 December 2024); WCAG-EM 2.0 (W3C Group Note, 23 July 2026)
ibm.com/able/product_accessibility/— IBM Product Accessibility Reports catalog, filtered by hand 27 July 2026- IBM SkillsBuild AI Certificate Syllabus and Course Outline, Skills Framework, and Learning Journey, all dated 07.24.2026
- Enablement Playbook for the IBM SkillsBuild Artificial Intelligence Certificate, IBM, accessed 27 July 2026