Fortnightly briefing Policy · safety · AI · education
The guardrails are moving.
School strategy must move too.
Regulation is reaching into product design. AI is moving into health, study and sensitive conversations. Evidence from children is testing whether the new safeguards work. This issue connects those changes—and turns 29 important reads into five leadership priorities.
The signal
Protection is becoming
part of the architecture.
This fortnight’s clearest message is that online safety can no longer sit at the edge of a school’s digital strategy. Age assurance, addictive design, AI companions, learner-facing assistants and data governance are converging.
It seems unnerving to us that the technical landscape and considerations are changing so rapidly that it is difficult to build the right strategy to both utilise these new technologies or guard against any issues that could arise from their use.
As with any major innovative shift in educational technology, understanding and evaluating before any implementation is key. I hope these curated articles help in broadening your oversight of this exciting phase and of course Orca are here to help you on the journey.
Policy is moving from illegal content towards age, features and patterns of use.
Measurement and young people’s voices reveal whether safeguards work in practice.
Manipulated imagery and networked offending demand earlier, clearer intervention.
Search, health and sensitive conversations raise fresh safeguarding questions.
Schools need impact measures, ethical criteria and coherent learner entitlement.
Rules are entering
the product.
The UK direction of travel is unmistakable: safety policy is moving into age assurance, default settings, persuasive design and data governance. Schools need to explain the change—and anticipate its operational consequences.
The Data (Use and Access) Act 2025: what schools need to know
Updated guidance explains amendments to UK GDPR, the Data Protection Act and related rules—directly relevant to AI procurement, learner data and automated tools.
Open the school guidance ↗Fact sheet: new rules to protect children online
The clearest current explanation of proposals covering an under-16 social-media ban, age checks, risky features and planned safeguards for AI chatbots.
Read the fact sheet ↗Online Safety Act: implementation collection
The central reference for commencement, enforcement and policy updates under the Online Safety Act 2023.
Track implementation ↗Growing up in the online world: progress statement
The evidence and political rationale behind the under-16 ban, age assurance, risky-feature controls and device measures.
Read the progress statement ↗2026–29
A Safe, Informed Digital Nation
The national Media Literacy Action Plan sets priorities for children, trusted information, local initiatives and cross-system coordination.
Explore the action plan ↗Measure what children
actually experience.
Regulatory progress matters, but it must be tested against children’s real use, exposure and reporting behaviour. These seven reads provide a stronger evidence base for school decisions.
Children’s online experiences research report
A comprehensive account of children’s service use and exposure to harmful content before and after the new safety duties. It is the strongest current UK evidence base for testing whether regulation is changing lived experience.
Read the Ofcom report ↗
Younger phone owners, the rise of AI and consumption over creation
An accessible synthesis covering earlier device use, generative AI and changing creative habits.
Explore the findings ↗8–14
Children’s Passive Online Measurement
Behavioural measurement of the websites and apps used by a UK-representative sample helps leaders challenge assumptions about where pupils spend time.
View the measurement data ↗early impact
Age checks helping make experiences safer—but the job is not done
Ofcom reports early progress while challenging platforms to strengthen protections and age inference.
Read the regulator update ↗and reporting
The Ember Project: what young people tell us about reporting online
Exploratory research captures young people’s understanding of online child sexual exploitation and the barriers that can stop reporting.
Read the Ember findings ↗record demand
Revenge Porn Helpline handles record number of cases in 2025
A sharp increase in cases and removal requests is a sobering signal for safeguarding, relationships education and routes to specialist support.
Read the helpline update ↗research ahead
Research on AI chatbots, social media and video-sharing platforms
Ofcom will use child-profile accounts to examine recommended feeds and safeguards in generative-AI chatbots.
Review the transparency notice ↗AI changes both
scale and proximity.
Emerging threats are not remote technical problems. They can begin with ordinary school photographs, everyday platforms or rapidly forming online networks.
AI-generated child sexual abuse material: new guidance for parents
IWF and NCA guidance explains how openly shared images of children are manipulated and what families can do. Schools should review image publication, parent communication and reporting routes.
Open the practical guidance ↗
Five Eyes leaders call for stronger action on serious online crime
Priorities include AI-enabled crime against children, sadistic online networks and growing offender capability.
Read the strategic warning ↗The boundary of AI
keeps expanding.
Search, emotional context, health, agents and more capable frontier models are changing what school policies must anticipate. Provider claims need to be read alongside independent scrutiny.
stress test
Google’s AI search fails child-safety tests
Independent testing raises concerns about the safety of AI answers already used by a large majority of US teens and tweens. For schools, this is as much a critical-literacy issue as a product-safety issue.
Read the investigation ↗Safety, governance and provider direction
policy position
Advancing AI safety through state and federal action
A primary-source view of OpenAI’s preferred framework, including youth protection and AI literacy.
Read OpenAI’s position ↗safety index
AI Safety Index — Summer 2026
An independent comparison of leading AI companies’ governance, risk management and safety commitments.
Open the index PDF ↗
Google AI announcements from June 2026
A compact horizon-scanning roundup from a supplier already embedded in many school environments.
Review the announcements ↗conversations
Helping ChatGPT recognise context in sensitive conversations
Changes aim to identify risk that becomes apparent over longer interactions—especially relevant when young people seek emotional support from general-purpose AI.
Explore the safety changes ↗Frontier releases and new product boundaries
Launching Health in ChatGPT
A US health-focused experience designed around personal health context, privacy controls and clinician input. Its arrival sharpens questions about pupils seeking health or wellbeing advice from AI.
Read the launch details ↗Sonnet 5
Introducing Claude Sonnet 5
A frontier model for coding, agents and professional work that may change capability assumptions behind school policy and staff development.
Review the model release ↗Move from access
to purposeful use.
The strongest implementation signal is not a new feature. It is the need to define impact, professional support, ethical boundaries and a coherent learner entitlement before scale.
Teaching & learning
Measuring the impact of AI on teaching and learning
Two studies report improved maths learning and more tailored lesson planning when Gemini is paired with educator support. Leaders should examine both the outcomes and the conditions behind them.
Read the impact studies ↗School-centric AI products
AI tailored for education, with educators in the lead
Gemini and Classroom features span guided learning, study notebooks, teacher workflows and administration.
Explore the educator updates ↗
Study and Learn: AI built for students
A learner-facing Microsoft 365 Education experience with administrator controls and education privacy claims.
Review Study and Learn ↗for students
Supporting students with connected AI tools
Gemini study notebooks and Classroom connections are planned for school-issued under-18 accounts—prompting review of age, data, safeguarding and classroom controls.
Explore the student updates ↗European digital education initiatives
competences
New AI Literacy Framework helps schools prepare learners
Classroom examples for primary and secondary education offer a strong international comparator for coherent AI literacy beyond tool training.
Explore the framework ↗toolkit ahead
Hub squad on ethical AI toolkit for schools
A practical European toolkit is in development, with publication planned for autumn 2026.
Follow the toolkit work ↗and trust
Digital education content: guidelines and interoperability framework
Guidelines address quality, safety, trustworthiness and inclusion when selecting, creating and using digital content.
Review the European guidance ↗Five practical moves
Turn the signal
into action.
Use the reading as a prompt for joined-up decisions—not as another list to circulate.
- 01
Connect safeguarding and data governance
Review AI procurement, learner data, age assumptions and incident response through one decision process.
- 02
Test policy against pupils’ lived experience
Use Ofcom’s evidence and pupil voice to challenge assumptions about use, exposure and reporting.
- 03
Review the school’s image footprint
Revisit publication, consent and parent guidance in light of AI-manipulated imagery.
- 04
Define boundaries for sensitive AI use
Make health, wellbeing, emotional reliance and escalation routes explicit in staff and pupil guidance.
- 05
Measure implementation, not adoption
Set success criteria for learning, workload, inclusion and professional judgement before expanding access.
Need a clearer next move?
Make the next decision
with confidence.
Orca Advisory helps schools and children’s professionals connect safeguarding, digital culture, AI strategy and whole-school improvement.
Policy signal
New social-media curfews and crackdown on addictive features
Default midnight-to-6am curfews and restrictions on addictive features are proposed for 16- and 17-year-olds. The policy moves beyond content rules into product design and time-of-use controls.
Read the announcement ↗