California school districts are employing AI in classrooms to ready students for tomorrow, but early experiments have already triggered controversy.
In December 2025, a fourth-grade class at an elementary school in the Los Angeles Unified School District (LAUSD) received a homework assignment asking students to create a book cover of Pippi Longstocking. One of the fourth-grade students asked the program to generate an image of “long stockings, red haired girl with braids sticking out.” But the image produced for the student was explicitly sexual.
The tool used by the students was Adobe Express for Education, a graphic design software program. Julie, An LAUSD parent, told the Daily Caller News Foundation that the program was downloaded through a learning management system called Schoology. Julie noted that Schoology is the district’s platform for its approved K-12 content.
It is unclear whether students were explicitly instructed to use the AI tool, but Julie stated they weren’t discouraged from using it.
In three of the four photos given to the student a woman in a tight short skirt with long black stockings could be seen standing in the picture. A fourth image showed the woman in what appears to be black lingerie, with stockings and heels.
“Other parents were able to replicate that on their kids’ Chromebooks. So they basically took the same prompt that the girl had entered, and they were also getting highly sexualized images. So it’s not like this was like an edge case, a one-off,” Julie told the DCNF.
“My take is that it feels really sneaky. Basically, Google Gemini just happened to be in the Google suites that all students with Chromebooks had access to, and that was never explicitly stated to any parents. I don’t even know if it was explicitly stated to any teachers. So it just suddenly was there,” Julie said.
In 2023, California schools began to discuss implementing artificial intelligence into classrooms after the rise in use of ChatGPT. The California Department of Education (CDE) later released an initial guidance in 2023 titled “Learning with AI, Learning about AI,” which focused on AI benefits and its ethical use in classrooms.
Structures and pilots rolled out the following year, and the CDE released its latest guidance just last year on the safe and effective use of AI in TK-12 schools, supporting districts in implementation with ongoing professional learning and resources. However, one school in the LAUSD, the second largest in the nation, recently came into the media spotlight for a tool used by fourth-grade students.
LAUSD did not respond to the DCNF’s request for comment.
That lack of transparency highlights broader issues in how California handles AI in education. The state’s local control model leaves most regulation to the districts. Christian Pinedo, vice president of external affairs and advocacy at The AI Education Project, who sits on the CDE’s AI working group, explained that this setup can lead to uneven safeguards and oversights.
“California is a local control state. Each individual school board creates their own policies and passes their own policies for each individual school district,” Pinedo said. “And so you’ve seen the California School Board Association put out policy recommendations and policy frameworks … but they cannot go past passing recommendations or frameworks.”
Julie also pointed to a district-wide “hour of AI” encouraged for K-12 students at the end of last school year, using a program through code.org to choreograph a dance. Julie opted her kindergartner out, but she questioned the consistency with LAUSD’s own policy requiring students to be 13 or older for generative AI use, along with digital citizenship training.
The training is a mandatory webinar LAUSD’s Division of Instruction launched for all schools with students TK-12. According to LAUSD’s site, elementary schools are instructed to implement Common Sense Education’s lessons, while secondary schools deliver the “Digital Citizenship in the Age of AI” lesson.
“This initiative aligns with LAUSD policies to ensure a safe and responsible digital learning environment,” the site states.
“Their stated policy on generative AI is that a student has to be 13 years or older to use it. And they have to have done digital citizenship training,” Julie told the DCNF. “So in that case there, you know, the policy exists, but I don’t know who enforces the policy. We’ve pointed this out to them [LAUSD] many times.”
“But then in the same breath, they’re promoting an hour of a district wide hour of AI for K through 12. So it’s kind of like have their cake and eat it, too,” Julie said.
Amy Eguchi, a teaching professor in computer science education at the University of California, San Diego’s Department of Education Studies, has been involved in AI for K-12 initiatives since around 2016. She stressed that elementary students should not use AI tools unsupervised, aligning with common age restrictions like those for social media.
“We don’t recommend elementary school teachers to let their students use AI tools without adult supervision,” Eguchi said.
Experts like Eguchi and Pinedo agree that without stronger guidelines, incidents like this are bound to happen, especially when tools rely on user prompts that can pull from biased or inappropriate internet data.
Eguchi explained that tools like Adobe’s Firefly, which powers the image generation in Express for Education, depend heavily on how prompts are worded, and unexpected outputs can stem from vague or incomplete instructions—a classic case of “garbage in, garbage out.”
“The quality of prompting is so important with the current generative AI,” Eguchi said. “So if you don’t specify the image, I’m guessing it probably didn’t ask like, ‘Oh, this is the picture book for fourth graders.’”
Eguchi also noted that some children are online outside the classroom and being exposed to similar images on social media.
“I was talking to the person who actually started the AI for K-12 initiative,” Amy told the DCNF. “And her take is, ‘Well, but this is like something that you see online all the time.’ The kids probably see something like that all the time. That kind of animation, right?”
“So that’s a very typical anime kind of image that kids are kind of exposed to,” Eguchi added. “Then also like K-pop. And if you come to Japan, that image is like all over the place. There’s no filter in a way for kids.”
But Julie pushed back hard on that notion, arguing that blaming the prompt shifts accountability onto the child and ignores AI’s inherent flaws. She noted the prompt was straightforward and innocent, yet still yielded inappropriate results, making it unfair to expect a fourth grader to craft it perfectly.
“Okay, here’s what the prompt was… long stockings, red haired girl with braids sticking out. So tell me what about that prompt asked for the result that it got. Was it the word girl? Was it the words long stockings? I just think it’s insane,” Julie told the DCNF. “So the nine year old who entered a prompt… It’s her fault? So somehow she did it wrong.”
Julie also emphasized that the responsibility shouldn’t fall on individual schools or teachers, who are already overstretched.
But some parents, like Julie, don’t agree with generative AI in schools.
“We know so little. So to me, it seems insane, the rush to put this in front of developing learning minds,” Julie said.
All content created by the Daily Caller News Foundation, an independent and nonpartisan newswire service, is available without charge to any legitimate news publisher that can provide a large audience. All republished articles must include our logo, our reporter’s byline and their DCNF affiliation. For any questions about our guidelines or partnering with us, please contact [email protected].
