Fortnightly briefing 30 August–13 September 2026
Keep human judgement
at the centre.
A school-year safeguarding reset. Better questions about AI and learning. And a measured look at the warnings coming from AI’s builders. Eighteen selected reads to help education leaders make the next decision with care.
Editor’s note01Give the decision
room to breathe.
The new school year brings immediate responsibilities. The AI industry brings increasingly urgent claims about the future. Both deserve judgement, and neither removes our responsibility to children now.
This issue starts with safeguarding arrangements, image abuse and what families and young people report. It then looks at AI in teaching and the current debate about catastrophic risk. We should listen carefully when developers call for stronger safeguards, while distinguishing personal forecasts from observed behaviour and measures already in place.
For schools, the practical response is to stay clear about purpose, evidence and accountability. Ask what a tool can access, who checks consequential decisions and how concerns are reported. Give staff space to learn and pupils space to think. Safeguarding and the wellbeing of children and young people should remain at the heart of every decision.
Start with the arrangements
children rely on.
New announcements matter when they lead to clear responsibilities and reliable practice. England-specific guidance and UK platform regulation have different scopes.
reset
A school-year safeguarding reset
September is a useful moment to reconnect policy, staff confidence and pupils’ reporting routes. SWGfL’s briefing brings online safety back into everyday school routines. Ask the safeguarding lead, IT lead and a governor to identify one arrangement that needs checking in practice this fortnight.
Read the source ↗monitoring
Filtering and monitoring: check the arrangements, not just the policy
The current DfE core standard sets out expectations for schools and colleges in England. The wider manual was updated on 8 September; that does not mean every provision changed that day. Review responsibilities, checks and incident records together, including the limits of filtering dynamically generated AI content.
Read the source ↗and health
Revised RSHE guidance is now in effect in England
The revised relationships, sex and health education guidance applies from 1 September 2026. This is an implementation milestone for guidance published earlier. Check curriculum planning, staff preparation and communication with families against the applicable guidance, including learning about online relationships and harm.
Read the source ↗accountability
Ofcom puts platform action on intimate-image abuse in focus
Ofcom’s 9 September announcement sets a 30 September deadline for relevant services to use hash matching, or equally effective measures, against illegal intimate imagery. These are platform responsibilities. For schools, the practical question is whether pupils know how to disclose image abuse and reach appropriate support.
Read the source ↗Listen before
you draw conclusions.
These August sources provide recent context for September decisions. Keep reported experience, confirmed incidents and evidence of impact distinct.
literacy
AI and literacy: widespread use needs better support
The National Literacy Trust’s August research draws on young people, teachers and parents. It highlights the need for critical engagement and support alongside AI use. Survey responses describe reported experience; they do not establish that AI improves learning. Use the findings to ask where staff development and pupil support are uneven.
Read the source ↗and disclosure
Manipulated images: take the disclosure seriously
The IWF reports that 420 under-18s using Report Remove in the first half of 2026 said their imagery was faked or manipulated, compared with 397 across 2025. These are reports of suspected manipulation, not confirmed AI cases. Rehearse a calm, non-blaming response and age-appropriate reporting routes.
Read the source ↗families
Parental confidence can conceal gaps in understanding
The ICO’s August research explores differences between parental confidence and children’s digital experience. Use it to open a conversation with families about what children actually encounter, where they seek help and what adults may miss. A familiar device or platform does not guarantee shared understanding.
Read the source ↗When AI’s builders ask
for stronger brakes.
The current “doomsday” dialogue deserves attention without treating catastrophe as a settled forecast. Three complementary voices raise questions about pace, oversight and accountability.
pace and safeguards
Anthropic’s chief executive calls for a slower frontier
In “We Must Pace the Frontier”, Dario Amodei argues that safeguards should constrain the pace of capability growth. He proposes embedded independent evaluators and international coordination. His warning is a judgement about possible future risks; the proposals should be assessed through their implementation and evidence.
Read the source ↗risk assessment
An incoming OpenAI Foundation board member challenges the industry
Paul Christiano’s 9 September personal statement questions whether the industry, including OpenAI, is doing enough to reduce loss-of-control risk. His forecasts are explicitly subjective. His appointment is to the Foundation board, with a non-voting observer role on the Group board; this is his personal assessment, not an agreed company forecast.
Read the source ↗mandatory safeguards
OpenAI argues for mandatory safety requirements
OpenAI’s policy statement calls for national, capability-based safety requirements in the United States. Read it alongside the personal warnings to distinguish advocacy, commitments and controls already in operation. It is a company policy position, not a new requirement for UK schools.
Read the source ↗Orca’s reading: schools do not need to settle the extinction-risk debate to demand clear access controls, human approval, incident reporting and independent evidence. This is an editorial implication, not a directive from either company.
Protect the thinking
behind the tool.
Start with the educational purpose. Parliamentary testimony, programme accounts and supplier announcements each offer a different kind of evidence.
and assessment
Parliament examines AI, teacher expertise and assessment
The Education Committee’s 8 September hearing brings academic, teaching and assessment perspectives into the discussion of AI and EdTech. These are witnesses’ accounts, not the committee’s final conclusions. Use the transcript to sharpen questions about teacher expertise, assessment validity and what evidence should justify wider adoption.
Read the source ↗AI literacy
Primary AI literacy: start with understanding
The National Literacy Trust’s account of a Portsmouth programme offers an August practice example for primary education. Use its approach to inform age-appropriate learning about AI and critical thinking. A programme account can inspire local planning; it does not establish that the same approach will deliver the same outcomes everywhere.
Read the source ↗learning materials
Ask what AI changes inside the learning material
EdReports’ September scan examines AI in K–12 instructional materials through a qualitative look at ten providers. This US market perspective offers useful procurement questions, rather than a comprehensive product ranking. Ask suppliers what changes for pupils, what teachers can control and what independent evidence supports claimed benefits.
Read the source ↗with purpose
A new educator training series: use it with a clear purpose
Google’s 3 September announcement introduces AI educator training. It offers a staff-development starting point from a technology supplier. Choose a specific teaching or workload question, examine the examples critically and agree how any tool use will fit your school’s safeguards and expectations.
Read the source ↗Make the next step
small and useful.
A policy review, a pupil conversation or focused staff learning can turn a busy news cycle into manageable school improvement.
against practice
Updated 360Safe and policy templates support review
SWGfL’s 26 August update covers its online-safety self-review tool and policy templates. Treat templates as a starting point for a locally owned review. Check that responsibilities, reporting routes and practice described in the document match what staff and pupils would actually experience.
Read the source ↗cyberbullying
#NotOnOurFeed: a timely cyberbullying conversation
The European school education platform highlights the campaign running to 25 September, with a webinar on 16 September and further training signposted. Use the resources to connect pupil voice, bystander behaviour and trusted reporting routes. Check registration and availability on the source page before planning attendance.
Read the source ↗learning
Self-paced courses for the next staff-development step
The European school education platform’s 7 September update signposts self-paced professional learning. Review course content and access before allocating staff time. Link participation to one improvement priority and a short follow-up discussion about what will change in classroom or safeguarding practice.
Read the source ↗purpose and practice
Phone bans: examine implementation as well as the headline
The European school education platform’s 31 August overview explores mobile-phone bans in schools. It offers a comparative perspective, not a single answer for every setting. Review your own approach against pupil experience, learning, wellbeing, accessibility and how consistently adults can implement it.
Read the source ↗Turn one question
into a clear action.
Give these three conversations an owner, a deadline and a way to check the result.
- 01
Check a safeguarding route
Walk through how a pupil reports image abuse or an upsetting AI interaction. Agree who responds and how support is coordinated.
- 02
Examine one AI use case
Name the learning purpose, the information the tool can access and the person accountable for checking its output. Decide what evidence would justify continuing.
- 03
Listen and follow up
Ask pupils, families and staff where current arrangements help and where they fall short. Bring one finding back to the next meeting.
Make the next decision
with confidence.
Orca Advisory helps schools and children’s professionals connect safeguarding, digital culture, AI strategy and whole-school improvement.