Fortnightly briefing 14–27 September 2026
Keep the conversation
open.
Policy is moving. AI capability is moving faster. Twenty-eight selected reads to help education leaders keep children’s experience, professional judgement and practical action in the same frame.
Editor’s note01Make room for
the human signal.
The rules around children’s online lives are being debated, refined and enforced. At the same time, the tools children and staff encounter are becoming more capable, more personalised and harder to explain.
This issue starts with the policy environment, but its central question is practical: can a child name an adult who will listen, respond and follow through? Proposed legal change and platform controls matter. They work best alongside confident reporting routes, clear family communication and staff who understand the experience behind a disclosure.
The AI material offers a similar lesson. A product release or a promising classroom example should lead to better questions, not automatic adoption. Be specific about the educational purpose, access to information, human checks and the evidence that would justify continuing. Keep safeguarding and the wellbeing of children and young people at the heart of every decision.
Keep the rules
in proportion.
Government proposals, statutory guidance and platform enforcement shape the wider landscape. They do not remove the need for calm, locally owned safeguarding practice.
briefing
The Social Media Ban: What Does It Mean for Schools and Professionals?
SWGfL's webinar briefing explains the proposed under-16 social-media ban, platform responsibility and the continued role of education, pastoral support and disclosure routes.
Read the source ↗Editorial note: proposed legislation and reported political responses are not automatically new school duties. Check the source, scope and commencement before changing local practice.
briefing
SWGfL explains proposed child online-safety legislation
SWGfL responds to plans for stronger protections concerning children's access to nude images.
Read the source ↗and accountability
Government prepares legislation on children's access to nude images
The government says it will introduce legislation requiring stronger device and platform safeguards to prevent under-18s taking, sharing or viewing nude images, addressing grooming, exploitation and abuse before exposure.
Read the source ↗and accountability
Information sharing duty statutory guidance
DfE's consultation outcome links to statutory guidance and templates for the new information-sharing duty, which applies from September 2026 to support safeguarding practice.
Read the source ↗and accountability
Mobile phones in schools
DfE statutory guidance says schools should implement a whole-school-day phone-use prohibition from 1 September 2026, aligned with behaviour, welfare and safeguarding duties.
Read the source ↗and accountability
Ofcom fine follows failure to introduce age checks
Ofcom reports a £730,000 fine and the subsequent introduction of age checks on an adult-content site.
Read the source ↗and scrutiny
Social media ban: Committee publishes government response to report on children’s screentime
The Education Committee publishes the government response to its social-media consultation report, including planned regulations and the position that online harms are a safeguarding and public-health issue.
Read the source ↗and scrutiny
Growing up in the online world: government response, July 2026
The consultation response explains the government's approach to social media restrictions, AI-related concerns and media literacy.
Read the source ↗Listen for what
children are navigating.
Research can sharpen a school’s questions without pretending to describe every community. Use national evidence alongside pupil voice, family conversations and your own safeguarding records.
perspective
Insights from the special panel on child safety online
The article brings together young people's and experts' reflections on protecting minors in digital environments.
Read the source ↗young people
Understanding online communications among children
Ofcom's 2026 UK study of 4,097 11–17-year-olds examines online communication, expanding networks, uncomfortable contact, sexualisation and encounters with content about drugs, weapons and terrorism.
Read the source ↗young people
Update: tech firms’ responses to our call for action to protect children
Ofcom's Project Mercury update records platform responses to its child-safety call for action, alongside evidence and questions about the effectiveness of measures.
Read the source ↗young people
ChatGPT in lesson preparation: Teacher Choices trial
An NFER-evaluated trial involving 259 teachers in 68 schools found lower KS3 science preparation time with ChatGPT and a supporting guide, without an apparent reduction in assessed resource quality.
Read the source ↗young people
Talking to machines: Children’s experiences with AI assistants and companions
A survey of 1,950 children aged 10–17 explores use of AI assistants and companions for learning, social and personal reasons, plus reported harmful interactions and sensitive-data sharing.
Read the source ↗Respond with care,
not panic.
Synthetic imagery, unsafe contact and model behaviour all demand knowledgeable adults and trusted routes to support. Clear disclosure, proportionate evidence and coordinated response remain the essentials.
briefing
The Law and Non-Consensual Synthetic Intimate Images (‘Deepfakes’)
SWGfL explains the legal position on creating, sharing and threatening to share non-consensual synthetic intimate images, alongside routes to specialist support.
Read the source ↗Editorial note: a calm, non-blaming response makes disclosure more likely. Follow safeguarding procedures and use specialist reporting/support routes where appropriate.
briefing
Preventing and responding to intimate-image abuse
SWGfL links Ofcom's intimate-image measures with prevention, support services and industry cooperation.
Read the source ↗briefing
Appropriate filtering and monitoring: what's new for 2026?
UKSIC explains the updated definitions, including real-world effectiveness, app coverage, AI-generated content and technical limitations.
Read the source ↗understand
National Strategic Assessment 2026: online harms
The NCA's online-harms assessment provides a law-enforcement perspective on the threats facing children and the wider public.
Read the source ↗understand
Our framework for reporting model misalignment
OpenAI introduces a framework for reporting unexpected or concerning model behaviour, alongside six reports from the previous six months, including cases that may be disclosed before full explanation or mitigation.
Read the source ↗understand
Supporting content creators affected by intimate-image abuse
SWGfL explains support for adult content creators whose consensually shared intimate material is subsequently abused.
Read the source ↗Read capability
with safeguards.
Rapid product announcements tell us what systems may soon enable. They are not evidence that a setting should adopt them. Keep access, human oversight and escalation routes visible.
inside
Research acceleration: The view inside OpenAI
OpenAI reports early internal measurements of coding agents contributing to research work and sets out its view of the safeguards and public transparency needed as capabilities advance.
Read the source ↗Orca’s reading: a capability announcement is a prompt for due diligence, not an adoption decision. This is an editorial implication, not a claim from the providers.
and safety
Safety overview: GPT-6 Astra
OpenAI describes Astra's capabilities, its Critical cybersecurity classification under its own framework, and safeguards associated with release.
Read the source ↗and safety
Introducing Claude Fable 5.1 and Claude Mythos 5.1
Anthropic announces models with different safeguard and access arrangements: general availability for Fable and trusted-access programmes for Mythos.
Read the source ↗and safety
Introducing Gemini 3.8 Flash and 3.8 Flash Cyber
Google introduces new Flash models, including a cybersecurity-focused variant with distinct access and safeguards.
Read the source ↗Make every pilot
answer a question.
Examples from education providers can help schools move beyond general enthusiasm. Start with the curriculum or workload problem, involve staff, and decide how you will test benefit, risk and equity.
implementation
AI in education: From experimentation to institutional impact
Microsoft argues for phased, institution-led adoption: begin with one workflow, use existing controls and expand only as governance and confidence mature.
Read the source ↗implementation
Giving teachers a starting point with curriculum-aligned AI tools
Teachers and leaders at Frome Vale Academy describe using Oak's AI tools to start planning, follow curriculum sequences and adapt resources.
Read the source ↗implementation
What's new in the Microsoft Education AI Toolkit
Microsoft's updated AI Toolkit groups planning, implementation, research and peer-learning resources for institutions moving from pilots to wider use.
Read the source ↗implementation
Teaching and learning with artificial intelligence
A school-practice account explores how learning about AI can include its social context, privacy, information security and responsible use.
Read the source ↗update
Google's Back to School 2026 education updates
Google brings together student study tools, classroom connections and education offers for the new academic year.
Read the source ↗Turn care into
clear practice.
Give these three conversations an owner, a deadline and a way to check the result.
- 01
Test a reporting route
Walk through how a pupil would disclose image abuse, an unsafe online contact or an unsettling AI interaction. Agree how the first adult responds and who coordinates support.
- 02
Check one proposed change
Separate what has been announced, consulted on or enforced from what currently applies to your setting. Make any policy response proportionate to the actual requirement.
- 03
Choose one AI use case
Name the teaching or workload problem, the data involved and the human accountable for checking the output. Decide what would count as a safe and useful result.
Make the next decision
with confidence.
Orca Advisory helps schools and children’s professionals connect safeguarding, digital culture, AI strategy and whole-school improvement.