Almost none of them. As of mid-2026, the most popular AI social story tools (MagicSchool, ChatGPT, Claude, Gemini) generate text only. The school SLP still hunts for images separately. From our 2024 community survey (n=16 parents, school SLPs, OTs, SPED teachers), the most repeated complaint was: "Getting suitable pictures is 90 percent of the work." A few 2026 tools (SOFA app, Emoquest private beta, Canva Magic Write paired with Canva's stock library) include illustrations or photos, but each has tradeoffs on cost, FERPA, and Carol Gray methodology fit.
Why do most AI social story tools skip pictures?
Because consistent multi-page illustrations are hard. A social story needs the same student or character across 4 to 8 pages, in different scenes, with the same hair, skin tone, and clothes. Generic image models (DALL-E, Midjourney, Stable Diffusion) drift from page to page. Specialized education tools opted out of images entirely rather than ship inconsistent ones. The result: the AI saves 5 minutes on text and the SLP still spends 25 minutes on Google Images.
Which 2026 AI tools actually include pictures with the story?
Here is the practical landscape a school SLP or pediatric OT can choose from today:
| Tool | Text? | Pictures? | Cost (2026) | FERPA note |
|---|---|---|---|---|
| MagicSchool AI Social Story Generator | Yes | No | $8 to $15/mo | DPA available, signs district agreements. |
| ChatGPT (free or Plus) | Yes | One-off via DALL-E (Plus); not multi-page consistent | Free or $20/mo | No district DPA on the consumer tier; do not enter PII. |
| Claude (free or Pro) | Yes | No | Free or $20/mo | Consumer tier; do not enter PII. |
| Canva Magic Write + stock library | Yes | Stock photos and illustrations, manual pairing | $15/mo (Canva Pro) | Canva for Education tier available for K-12; check district approval. |
| SOFA (Stories Online For Autism) | Yes | Yes, simple illustrations | Free | Research app; check current data policy. |
| Pictello (iOS) | You write | You supply every photo | $19.99 one-time | On-device; FERPA depends on photo source. |
| Boardmaker (Tobii Dynavox) | You write | 80,000+ PCS symbols, not AI-generated | $5 to $15/mo individual | School-grade tool; established district approvals. |
| Emoquest (private beta) | Yes | Yes, AI-illustrated, K-5 styled | Free in beta | No real student PII required; illustrations only. |
Which tools are "AI text + you paste images"?
The honest majority. MagicSchool, ChatGPT, Claude, and Gemini all fall here. The workflow:
- Generate the text in the AI tool (1 to 3 minutes).
- Open a separate window, search Google Images or a paid stock site (10 to 25 minutes).
- Drop images into a Google Slides or PowerPoint template (5 to 10 minutes).
- Audit the Carol Gray descriptive-to-directive ratio (2 minutes).
Total: 20 to 40 minutes, which matches the 94% in our 2024 survey who said they spend 30+ minutes per story. The text-only AI helps a little; it does not solve the bottleneck.
Are AI-generated illustrations FERPA-safe?
Yes, as long as the illustration does not depict a real, identifiable student. An AI-generated cartoon kid getting a haircut is not a student record under FERPA, because no real student is identifiable from the image. A real photo of your student in the salon chair is a record, and needs written consent and district-approved storage. The simplest FERPA-safe pattern for an SLP is: AI-generated illustration + the student's first name only in the caption.
From our 2024 community survey: "I wish I had a template I could easily customize to change the pictures of the child or parents quickly but keep the same story." Text-only AI does not deliver this. A tool that swaps a generic illustration for the same scene with a different student's hair or skin tone is closer to the SLP's actual wish.
What does the research say about pictures vs text?
Two relevant findings. The Camilleri et al. 2023 SOFA digital social stories study (n=856), the largest digital social stories trial, found illustrated digital stories drove the strongest effects for younger and more verbal autistic students. The 2026 Frontiers in Psychology meta-analysis of 21 studies found digital formats had slightly higher effects than printed ones, though the difference was not statistically significant. Translation: pictures matter, but consistency and re-reading matter more than format.
How should a school SLP pick?
A decision tree that maps to your situation:
- District approved MagicSchool already? Use it for text + Canva for Education for illustrations. FERPA covered.
- No district AI approval yet? Use a 6-page Google Slides template + free stock photos (Pixabay, Pexels). No PII leaves the district drive.
- Need real photos of the student? Pictello on a district iPad with parent consent and district photo policy alignment.
- Want one tool that does text + illustrations? SOFA (free, research-grade), or join the Emoquest private beta if you serve K-5 autistic students.
What about AI tools that promise to generate everything?
A note on hype. Several 2025 and 2026 AI tools claim end-to-end social story generation with consistent illustrations. Most produce: visually inconsistent characters (page 1 hair color does not match page 4), generic stock-looking scenes that ignore the school context, or text that violates Carol Gray ratios (too many directive sentences). Always run the output through a Carol Gray methodology check before sharing with a parent. The AI is a draft, not a finished story.
Frequently Asked Questions
Which AI social story generator actually includes pictures?
Most AI social story tools (MagicSchool, ChatGPT, Claude, Gemini) generate text only. Tools that include illustrations as of 2026 are SOFA (research app, free), Emoquest (private beta, AI-illustrated K-5), and consumer tools like Canva Magic Write paired with Canva's stock photo library. Pictello generates layout but requires you to supply every photo.
Why does MagicSchool not include pictures with the social story generator?
MagicSchool is text-only by design across its 80+ educator tools. The social story generator outputs a sentence-by-sentence narrative; the SLP or teacher pairs it with images from a separate source (Google Images, the school's photo library, or a paid stock site).
Can I use ChatGPT to generate both the text and an image?
ChatGPT Plus can generate one image at a time via DALL-E, but it is not built for multi-page consistent characters and cannot reliably produce a 6-page story where the same student appears in each illustration. For a single static image (a setting or symbol) it works. For a full illustrated story it is impractical.
Are AI-generated illustrations FERPA-safe?
Yes, when the illustration does not depict a real student. AI-generated stock-style art (cartoon kids, classroom scenes from behind) is not a student record. Real photos of a student are FERPA records and need consent and district-approved storage. The simplest FERPA-safe option is AI-generated illustration plus a generic first name.
How much do these AI generators cost?
MagicSchool teacher suite: ~$8 to $15 per month. ChatGPT Plus: $20 per month. Claude Pro: $20 per month. Canva Pro: $15 per month. SOFA: free (research). Emoquest: private beta, free for early users. Boardmaker (Tobii Dynavox): $5 to $15 per month for individual, more for district licenses.
Why are pictures the bottleneck, not the text?
In our 2024 community survey (n=16), school SLPs, OTs, parents, and SPED teachers told us: getting suitable pictures is 90 percent of the work. Text takes a few minutes once you have a template. Finding scenario-appropriate, FERPA-safe images that match the student's grade level is the time sink no text-only AI tool solves.
Do illustrations need to look photo-real for a K-5 student?
No. Flat illustrations that show a student from behind or in profile (no face) generalize well for most K-5 students. K-2 students benefit slightly more from real photos when available, but a clean flat illustration outperforms cluttered clip art for almost every age.
One approach for a school SLP short on time and short on illustrations: keep a 5-tool stack. A 6-page Google Slides template, a folder of free stock photos sorted by scenario, a Carol Gray ratio cheat sheet, an AI text drafter (ChatGPT, Claude, MagicSchool, or Emoquest for one-sentence-in illustrated story output), and a delivery format your district already uses (Google Slides or PDF). The bottleneck is pictures, not text, so the highest leverage move is the image folder, not the AI tool.