The Department for Education (DfE) has published the 2025 edition of Keeping Children Safe in Education (KCSIE), its statutory safeguarding guidance for all schools and colleges in England. The revised guidance, currently issued as a draft for information, is due to come into effect on September 1, 2025.

The update introduces several new references and expectations related to digital safeguarding, reflecting evolving risks around artificial intelligence, cyber-security, and online misinformation. These changes appear in Part Two of the document and are directed primarily at governing bodies, proprietors, and designated safeguarding leads.

Misinformation and conspiracy theories formally classified as online harms

Paragraph 135 of the 2025 guidance expands the DfE’s definition of harmful online content to include misinformation, disinformation, and conspiracy theories. These are now listed alongside existing risks such as pornography, racism, radicalization, self-harm, extremism, and online grooming.

The section emphasizes that technology is a significant component in many safeguarding and wellbeing issues and that children may be exposed to multiple risks simultaneously, both online and offline. The explicit addition of misleading and conspiratorial content signals a growing concern about its impact on children’s development, decision-making, and exposure to extremist ideas.

New guidance on AI use in education

Paragraph 143 introduces a direct link to the DfE’s product safety expectations for generative artificial intelligence in schools. While the KCSIE document itself does not prescribe how AI should be used, it highlights the need for appropriate filtering and monitoring systems when AI tools are accessible to students.

This addition aligns with broader departmental efforts to balance innovation in digital learning with safeguarding and data protection obligations.

Self-assessment tools for filtering and monitoring

In paragraph 142, the DfE recommends that schools and colleges use the ‘Plan Technology for Your School’ tool, an online resource that allows institutions to self-assess their filtering and monitoring infrastructure. The tool supports compliance with the DfE’s filtering and monitoring standards, which require schools to:

• Identify and assign roles for managing digital safety systems

• Review provisions annually

• Block harmful content without disrupting teaching and learning

• Implement effective monitoring strategies appropriate to their safeguarding needs

Cybersecurity standards added to support digital resilience

Paragraph 144 refers to the DfE’s cybersecurity standards for schools and colleges, which were developed in collaboration with the National Cyber Security Centre (NCSC). These standards outline the technical and procedural steps education providers should take to protect systems, data, and users from cyberattacks.

Recommended actions include regular backups, access control policies, secure configuration of devices and software, and procedures for responding to data breaches. The DfE urges institutions to periodically review their systems to ensure they remain resilient to emerging cyber threats.

Annual online safety review recommended

While not new, paragraph 145 reiterates the expectation that schools and colleges conduct an annual review of their online safety provision. This includes carrying out a risk assessment that reflects the specific threats facing their student population, especially those who are considered more vulnerable.

The guidance also points to free self-review tools such as 360safe and the LGfL online safety audit.

Additional changes unrelated to technology

While the 2025 update is primarily technical, it also includes broader safeguarding revisions such as:

• Support for kinship care added to the role of Virtual School Heads (para 199)

• Statutory status for attendance guidance (para 177)

• Revised terminology aligned with the SEND Code of Practice (para 205)

• Signposting to new RSHE and gender-questioning children guidance (paras 128, 204)

• New resources such as Shore Space from the Lucy Faithfull Foundation (para 545) and safeguarding materials from the CSA Centre and The Children’s Society (Annex B)

When Sarah Leonardi filed to run for Florida’s Broward County School Board in 2019, she had no idea what she was getting into.

Leonardi won and took office in late 2020 in the middle of the pandemic. It was tumultuous. Gov. Ron DeSantis threatened to withhold school funding after the board defied his masking ban. Angry over mask mandates, some parents sent vitriolic emails and made veiled threats.

But as COVID rates began to ebb, new flashpoints emerged. In the fall of 2021, Leonardi chaperoned an elementary school field trip to a local bar and grill that happened to be gay-owned. Some conservative media ran with the story. New threats poured in.

“Some of them were like ‘You can’t outrun my Glock 9mm gun’ [and] ‘Take a dirt nap,’ ” Leonardi recalled in an interview with NPR. “One was like, ‘Sell that b**** as a sex slave to ISIS,’ which was oddly specific.”

Leonardi says she still receives threats when conservative media occasionally republishes the school field trip story.

“I’ll get an email or a phone call about it, just telling me what a horrific person I am,” she says.

Harassment and threats up 170% 

Leonardi’s experience captures how threats against local school officials across the U.S. have shifted and grown, according to researchers at Princeton University. They conducted what they say is the largest and most comprehensive study of its kind in the country. Princeton’s Bridging Divides Initiative interviewed Leonardi along with 38 other school board officials. They also surveyed more than 820 school board officials with a group called CivicPulse. Using open-source material, investigators documented threats and harassment against school officials from November 2022 through April 2023, and the same period two years later. They found such incidents rose by 170%.

Bridging Divides says some of the local cases corresponded with national attacks on diversity, equity and inclusion initiatives as well as on LGBTQ+ policies. Roudabeh Kishi, the project’s chief research officer, says the targets held a variety of political views.

“This isn’t really like a partisan issue,” she says. “We’re seeing really similar reports of experiences (on) all sides of the political spectrum.”

In addition to Leonardi, NPR interviewed six other current or former school board officials who said they had been targets of harassment or threats. They said the anger and distrust that developed during the pandemic helped fuel and shape future disputes over cultural issues.

“The pandemic started this conversation about what are individual freedoms,” says Alexandria Ayala, a former school board member in Florida’s Palm Beach County. “What can a government tell me to do or not do?”

A second “Civil War” in Gettysburg

Al Moyer, who’s now in his ninth year on the Gettysburg Area School Board in Pennsylvania, says battles over masking frayed relationships in the district. Then, in 2023, some people in the community became uncomfortable with a tennis coach who was transitioning to female and had used the girls locker room.

Moyer said one resident called a Republican board member who opposed renewing the coach’s contract a “Nazi” to her face. He says his wife lost friends over the controversy.

“Those two situations really caused a kind of second Civil War battle in Gettysburg,” Moyer says. “It was pretty ugly.”

School board members have to navigate fights over genuine issues, but increasingly they have to grapple with fake ones as well. Russell Devorsky, who recently retired after 14 years on a school board in suburban Waco, Texas, says false stories on social media sow confusion and fuel harassment. “I am consistently and constantly harangued with individuals saying, ‘Well, kids are dressing up like cats, and they have litter boxes in bathrooms,’ ” says Devorsky. “Even though there’s never been a school district that had that situation, people believe it.”

“Like pushing a wet rope up a hill”

Even ordinary issues — such as the construction of a new band hall — can be targets of misinformation, Devorksy says. He says there were false claims on social media that the hall wouldn’t be ready on time and that students wouldn’t have instruments. Trying to set people straight who consider comments on Facebook community pages authoritative is exhausting, Devorsky says. “It’s kind of like pushing a wet rope up a hill,” he says.

The Princeton researchers worry that harassment could drive some school board members to leave public service — which they are monitoring — or avoid engaging on controversial topics. But Sarah Leonardi, the one who took the students to the gay-owned restaurant, says she isn’t quitting because she feels like she’s still making a difference.

“Ultimately, I decided to move forward and run again,” Leonardi says. “That is just a sacrifice — or a vulnerability — I’m willing to accept for now.”

Patrick Drumm, principal of Auckland’s Mt Albert Grammar, said that with AI-models improving every month, schools were struggling to keep up.

Most teachers had studied traditional teaching techniques and now needed expert guidance on how to protect assessment integrity and fairness in the new age.

“You think you can set up a policy around [AI], but it’s gone by lunchtime,” he said.

The 876 exam breach investigations included 206 related to “authenticity”, which former Secondary Principals’ Association president Vaughan Couillault believed likely involved some use of AI.

Another 13 cases involved students with exams that had a “text increase in short time period”, presumably meaning they were accused of copying and pasting from somewhere else.

Seventy-one cases involved “navigating away from the digital platform”, suggesting that even opening a new window on a computer could offer a momentary but powerful way to cheat.

Drumm said AI’s “brave new world” left schools grappling with ethical questions and technology predictions that wider society hadn’t even fully addressed.

If humans come to rely on AI to do most of their writing in the future, for example, Drumm questioned what skills were most important for students to be learning now.

For instance, should they be assessed on how well they could write an essay themselves or on the quality of work they could get AI to prepare for them, he asked.

And how much change was needed before AI-generated brainstorming ideas were no longer considered plagiarised?

“What degree of modification is required … when does it become someone else’s work versus your own?” Drumm said.

On the flip side, getting AI to prepare high-quality work likely required good foundations in comprehension and logic that were typically best built using traditional methods like reading and handwriting, Drumm said.

“We’ve had really good evidence around … the whole process of the physical mechanics of writing being huge in terms of the cognitive process.”

Despite the challenges, NZQA deputy chief executive Jann Marshall said there were hundreds of thousands of entries in the country’s NCEA and NZ Scholarship each year.

The vast majority of students worked hard and fully complied with all rules, she said.

Secondary students needed to understand AI ahead of a future where it “will be an integral part of many industries”.

Her team were aware it brought opportunities and risks, she said.

They had created guidelines to help teachers by encouraging them to regularly check-in and observe progress on essays and other tasks, to chat to students about their conclusions and seek “careful” assistance from AI detectors.

When it came to exams overseen by the NZQA, the group did not allow the use of generative AI at all, Marshall said.

NZQA also had a range of other measures in place to “identify attempted breaches of the rules”.

Marshall said she couldn’t reveal them for security reasons.

Earlier this year, she said the aim was to have students use AI “appropriately to brainstorm information, not replicate or simply transfer the information into an assessment“.

She said 418 of the 876 breaches investigated resulted in action being taken.

Sometimes students were issued warning letters but still received their results.

“So for example, if a student takes notes into the assessment, but the notes wouldn’t help them in the assessment, then that’s often a warning letter.”

If the notes gave them an advantage, however, their result was withheld.