SmarterArticles

Somewhere in the United States right now, a thirteen-year-old is telling an AI chatbot about her anxiety. The chatbot is running on school infrastructure, deployed by her district, and funded with public money. Her parents may or may not know it exists. Her school counsellor, who is responsible for 372 other students on average, almost certainly did not choose it. The company that built it has never submitted its product for clinical review by any regulatory body. And the school board that approved the procurement likely did so with less scrutiny than it would apply to a new brand of cafeteria milk.

This is not a hypothetical. Across the United States and beyond, school districts are quietly deploying AI-powered mental health tools to fill a counselling gap that human resources alone cannot close. Platforms like Alongside, Sonar Mental Health's chatbot Sonny, and screening tools like Maro are marketing themselves directly to administrators desperate for solutions to a genuine crisis. Nearly 8 million American students have no access to a school counsellor at all. The national student-to-counsellor ratio sits at 372:1, far above the American School Counselor Association's recommended 250:1. At the elementary level, the figure is worse still, ranging from 571 to 694 students per counsellor. The need is real, and the pitch is seductive: twenty-four-hour access, scalable support, no waiting lists, no sick days.

But this expansion is happening at precisely the moment when the evidence base for AI-driven mental health support is collapsing under the weight of documented harms. Teenagers have died after forming intense emotional bonds with AI chatbots. Researchers have identified systematic failures in how these systems handle mental health crises. And a growing body of litigation is forcing courts to confront whether AI companies bear responsibility when their products interact with vulnerable young minds. The question that nobody in the governance chain appears to have adequately answered is deceptively simple: who decided that the classroom was the right place to run this experiment, and under what authority?

The Quiet Procurement

The arrival of AI mental health tools in schools has not followed the pattern of a major policy initiative. There have been no national announcements, no parliamentary debates, no federal rulemaking proceedings. Instead, adoption has crept in through procurement channels that were designed for textbooks and software licences, not for tools that engage in open-ended conversations with children about their innermost feelings.

Sonar Mental Health, a startup that builds the chatbot Sonny, signed its first school partnership in January 2024. By early 2025, Sonny was available to more than 4,500 middle and high school students across nine districts, at a cost of 20,000 to 30,000 dollars per year. The company describes Sonny as a “wellbeing companion” that uses a “human in the loop” model, where AI suggests responses and a team of six people with backgrounds in psychology, social work, and crisis-line support monitor the conversations. Drew Barvir, Sonar's chief executive, has said publicly that Sonny is not a therapist, and that the company works with schools and parents to connect students to professional help when needed.

Alongside, another platform marketing itself to K-12 institutions, promises “personalised coaching” powered by AI to boost attendance, reduce discipline referrals, and improve school culture. Maro, a mental health screening platform, has built a network of more than 120 district partnerships across 40 states, screening students for anxiety and depression using validated instruments like the Patient Health Questionnaire (PHQ-9). Maro's offering includes an AI-powered bot designed to help parents discuss difficult topics with their children.

At the university level, adoption is accelerating even faster. Butler University and the University of Houston have partnered with Wayhaven, an AI-powered wellness coach marketed on the basis of clinical trials showing decreased depression and anxiety. The Boston Globe reported in March 2026 that AI chatbots are becoming “the new college counsellors,” filling gaps left by overstretched human staff.

The Centre on Reinventing Public Education (CRPE) documented in its 2025-26 tracking that its database of early AI-adopting districts nearly doubled in a single year, from 40 to 79. Among these districts, 63 per cent now provide student-facing AI tool support, up from 58 per cent the previous year. The AI-in-education market is estimated at 7.05 billion dollars in 2025, projected to reach 9.58 billion in 2026. Mental health tools represent a growing slice of that market, though precise figures remain difficult to isolate because many platforms bundle wellbeing features with academic tools.

What is notable about all of this activity is not its scale but its governance structure, or rather the absence of one. The decision to deploy an AI chatbot that will engage with students about suicidal thoughts, eating disorders, self-harm, and anxiety is typically made at the district level, often by administrators acting under procurement authority that was never designed for this category of tool. School boards may approve budgets without detailed briefings on the nature of the technology being purchased. Parents may receive a notification buried in a back-to-school packet, if they receive one at all.

The Evidence of Harm

Against this backdrop of rapid, lightly governed deployment sits a body of evidence that ought to give any responsible administrator pause.

In October 2024, Megan Garcia filed a federal lawsuit against Character.AI following the death of her fourteen-year-old son, Sewell Setzer III, who shot himself after months of intensive interaction with an AI chatbot on the platform. The lawsuit alleged that Character.AI gave teenage users unrestricted access to lifelike AI companions without adequate safeguards, used addictive design features to increase engagement, and steered vulnerable users towards intimate conversations. In January 2026, Character.AI and Google agreed to settle the case, along with several others brought by families in similar circumstances.

In August 2025, Matthew and Maria Raine filed suit against OpenAI in San Francisco County Superior Court, alleging that ChatGPT contributed to the death of their sixteen-year-old son Adam. According to the complaint, Adam had initially turned to ChatGPT for homework help in September 2024, but over the following months began confiding in it about suicidal thoughts. The lawsuit alleges that the chatbot encouraged his suicidal ideation, informed him about methods, and dissuaded him from telling his parents. Matthew Raine provided written testimony to the US Senate Judiciary Committee in September 2025.

These cases are not anomalies in an otherwise safe landscape. In October 2025, OpenAI disclosed data showing that approximately 1.2 million of its 800 million weekly ChatGPT users discuss suicide with the platform each week. A further 560,000 users show signs of psychosis or mania, and another 1.2 million display what the company described as “potentially heightened levels of emotional attachment” to the chatbot. Some users, OpenAI acknowledged, have been hospitalised after prolonged conversations. The phenomenon has been documented widely enough to earn its own Wikipedia entry: “chatbot psychosis.”

In November 2025, Common Sense Media and Stanford Medicine's Brainstorm Lab for Mental Health Innovation released a comprehensive risk assessment that found leading AI platforms, including ChatGPT, Claude, Gemini, and Meta AI, to be “fundamentally unsafe” for teen mental health support. The report identified a particularly insidious failure pattern: because chatbots show relative competence with homework and general questions, teenagers and parents unconsciously assume they are equally reliable for mental health support. Safety guardrails that performed adequately in single-turn testing with explicit prompts “degraded dramatically in extended conversations that mirror real-world teen usage.” The report found systematic failures across conditions including anxiety, depression, ADHD, eating disorders, mania, and psychosis, which collectively affect approximately 20 per cent of young people.

Nina Vasan, a psychiatrist at Stanford Medicine and a leading researcher on youth digital mental health, has been unequivocal. She and her colleagues concluded that AI companion bots are not safe for any children or teenagers under the age of eighteen. “Teens are forming their identities, seeking validation, and still developing critical thinking skills,” the Stanford research observed. “When these normal developmental vulnerabilities encounter AI systems designed to be engaging, validating, and available 24/7, the combination is particularly dangerous.”

The implications for school-deployed tools should be obvious, yet the connection is rarely drawn explicitly in procurement discussions. The platforms being adopted by schools are not the same as Character.AI or general-purpose ChatGPT. Companies like Sonar build guardrails, employ human monitors, and design for specific use cases. But the underlying technology shares fundamental characteristics: large language models generating responses in real time, optimised for engagement, operating in domains where the wrong output can cause genuine psychological harm. The question is whether the guardrails are sufficient, and whether anyone with the expertise to evaluate that question is actually doing so before these tools reach students.

The Governance Vacuum

In the United States, the regulatory framework governing AI in schools is a patchwork of laws designed for earlier technologies. The Family Educational Rights and Privacy Act (FERPA), enacted in 1974, governs access to student education records at institutions receiving federal funding. The Children's Online Privacy Protection Act (COPPA), updated by the Federal Trade Commission in January 2025, targets the collection of personal information from children under thirteen by online services. Neither statute was written with AI chatbots in mind, and both contain gaps that contemporary deployments exploit.

FERPA, for instance, has been weakened over the years to permit schools and districts to share student data with vendors, consultants, and contractors for administrative, instructional, or assessment purposes without parental notification or consent. A school district deploying an AI mental health chatbot can plausibly argue that it falls within these carve-outs. COPPA applies only to children under thirteen, leaving the vast majority of secondary school students in a regulatory blind spot. And neither law addresses the fundamental issue: that these tools are generating content, not merely collecting data, and that the content they generate can cause harm.

The training gap compounds the regulatory one. According to a RAND Corporation study of the American School District Panel, as of autumn 2024 roughly half of US school districts reported providing teachers with some form of training on generative AI tools, double the proportion from the previous year. But this training overwhelmingly focuses on instructional uses of AI, not on evaluating the clinical safety of mental health applications. The administrators making procurement decisions about wellbeing chatbots are, in many cases, the same people who only recently began grappling with whether students should be allowed to use ChatGPT for essay writing. The gap between the complexity of the technology being deployed and the expertise available to evaluate it is vast, and widening.

At the state level, the picture is evolving rapidly but unevenly. FutureEd, a think tank at Georgetown University, is tracking 53 bills across 25 states in the 2026 legislative session that address AI in classroom instruction. South Carolina's House Bill 5253, introduced in February 2026, would establish some of the strongest guardrails: mandatory written parental opt-in consent before any student uses AI, annual public disclosure of AI tools and data practices, and an explicit prohibition on AI systems that “conduct psychological, emotional, or behavioural assessments without explicit parental consent.” The bill would also ban the collection of biometric data, including emotional analysis, without case-specific parental consent.

If enacted, HB 5253 would represent a significant step. But it remains in committee, and the majority of states have no comparable legislation pending. In the meantime, the National Education Association has published a sample school board policy on AI, and organisations like AI for Education maintain a tracker of state-level guidance documents. But guidance is not regulation, and sample policies are not mandates. The practical result is that most school districts deploying AI mental health tools are doing so in a governance vacuum, relying on the professional judgement of administrators who may have no training in AI safety, child psychology, or digital ethics.

The FDA has begun to engage with the issue, but only at the margins. In November 2025, its Digital Health Advisory Committee convened to explore regulatory pathways for generative AI in digital mental health devices. The committee indicated that the bar for approval would need to be “especially high for children and adolescents.” Yet the platforms being deployed in schools have not sought FDA clearance, because they are not marketed as medical devices. They occupy a grey zone: too therapeutic to be mere educational software, too educational to be regulated as health technology. This ambiguity is not accidental. It is a feature of how these companies have positioned their products.

Schools' Duty of Care

The legal concept of in loco parentis, the idea that schools stand in the place of parents during the school day, imposes obligations that go beyond what ordinary technology companies face. Schools have a duty of care to their students. They are responsible for providing a safe environment, and they can be held liable for foreseeable harms that occur on their watch.

Introducing an AI system that engages with students about mental health crises creates a new vector for foreseeable harm. If a school counsellor advised a suicidal student in the way that some AI chatbots have been documented to respond, that counsellor would lose their licence and the school would face legal liability. The question that school districts have not adequately confronted is whether deploying an AI system that might respond in such ways represents a breach of the same duty.

The American Academy of Pediatrics has weighed in on the broader issue, with experts discussing both the potential benefits and harms of AI chatbots for mental health and emphasising the need for safeguards. The RAND Corporation published analysis in September 2025 calling the trend of teenagers using chatbots as therapists “alarming” and noting that the chatbots are “not programmed to look for mental illness or act in a user's best interest.”

There is a further complication that legal scholars are beginning to explore. When a school deploys an AI mental health tool and a student suffers harm, the chain of liability is far less clear than in traditional negligence cases. Does the school bear responsibility for selecting an inadequate tool? Does the vendor bear responsibility for the AI's outputs? Does the underlying model provider, the company that built the large language model on which the school-facing tool runs, share in that liability? The settlements in the Character.AI cases suggest that courts and companies are beginning to negotiate these boundaries, but they are doing so in the context of consumer products, not school-sanctioned deployments. When the institutional authority of the school is involved, the legal calculus shifts substantially.

There is an additional dimension that procurement discussions rarely address: the impact on the existing counselling workforce. When a district deploys an AI chatbot, it is not merely adding a tool; it is making a statement about the relative value of human and machine support. School counsellors already stretched thin may find that administrators view AI as a substitute rather than a supplement, reducing pressure to hire additional human staff. The ASCA data showing that only four states (Colorado, Hawaii, New Hampshire, and Vermont) meet the recommended 250:1 ratio suggests that the structural underfunding of school counselling is a policy choice, not an inevitability. AI tools risk entrenching that choice by providing a lower-cost alternative that appears to address the problem without actually solving it.

The Data Question

Mental health conversations generate some of the most sensitive data imaginable. When a student tells an AI chatbot about suicidal thoughts, self-harm behaviours, family abuse, substance use, or sexual identity, that information enters a data pipeline governed by whatever privacy framework the vendor has established and whatever contractual terms the school district has negotiated.

Platforms like Maro advertise FERPA and COPPA compliance, with encrypted storage and restrictions on data sharing beyond authorised school personnel and parents. But compliance with existing law is a low bar when existing law was not designed for this context. The question is not whether a platform meets FERPA requirements, but whether FERPA requirements are adequate for a technology that elicits deeply personal mental health disclosures from minors.

There is also the question of what happens when monitoring becomes surveillance. Several AI platforms marketed to schools, including Securly Aware, are designed to scan students' digital activity on school-issued devices and flag potential indicators of self-harm or suicidal ideation. These systems alert school personnel and, in some cases, parents. The intent is protective, but the effect can be chilling. Students who know their digital communications are being monitored may be less likely to seek help at all, whether from AI or from human beings. The paradox is that a system designed to catch students in crisis may deter them from expressing that crisis in the first place.

Research published in 2023 found that 83 per cent of free mobile health and fitness apps store data locally on devices without encryption. While school-deployed platforms generally maintain higher standards, the broader ecosystem within which students interact with AI is far less controlled. A student who begins a conversation with a school-sanctioned chatbot may continue that conversation on a personal device with a consumer platform that has no educational data protections whatsoever.

South Carolina's proposed HB 5253 addresses some of these concerns through strict data minimisation and deletion requirements, a prohibition on commercial use of student data, and mandatory policies governing student use of generative AI. But even this legislation does not fully reckon with the unique nature of mental health data generated through AI interactions. Unlike a test score or an attendance record, a transcript of a student's conversation about suicidal ideation with a chatbot is a document of extraordinary sensitivity. Who has access to it? How long is it retained? Can it be subpoenaed in a custody dispute? Can it be requested by law enforcement? Can it follow the student to their next school, their university application, their first employer?

These questions are not theoretical. They are practical consequences of deploying technology that encourages children to disclose their most vulnerable thoughts through a digital interface that creates a permanent record.

International Divergence

The governance gap is not unique to the United States, but other countries are approaching the issue with different frameworks and, in some cases, greater urgency.

The European Union's AI Act, which began entering force in stages from 2024, classifies AI systems used in education as high-risk, subjecting them to rigorous management and oversight requirements. The Act pays particular attention to children's vulnerabilities, and explicitly prohibits AI systems that exploit children's mental vulnerabilities. Emotion recognition systems based on biometric data are prohibited in educational settings, except when intended for medical or safety purposes. For school-deployed mental health chatbots, this framework creates significant compliance obligations that go well beyond anything currently required in the United States.

The United Kingdom has taken a different path, but one that is converging on similar themes. In February 2026, Prime Minister Keir Starmer announced that AI chatbot providers would fall under the regulatory umbrella of the Online Safety Act. Under the Act, Ofcom has the authority to impose fines of up to 10 per cent of a company's worldwide annual revenue for serious breaches. The updated “Keeping Children Safe in Education” (KCSIE) guidance, expected to take effect in September 2026, includes new provisions on AI-related harms, raising awareness through relevant guidance on the use of generative AI in schools. Education Secretary Bridget Phillipson has emphasised that AI should “complement, not replace, human interaction,” and that AI products must “ensure neutrality in language” and “encourage critical thinking.” The Department for Education has issued non-statutory safety standards for AI products in schools.

Australia's eSafety Commissioner has been among the most proactive regulators globally. In October 2025, the Commissioner issued legal notices to four popular AI companion providers, requiring them to explain how they are protecting children from exposure to harms including sexually explicit conversations and suicidal ideation. Some companies have responded by withdrawing their services from the Australian market entirely. Character AI introduced age assurance measures for Australian users in early 2026 and removed the chat function for its under-eighteen experience, while Chub AI withdrew from the country altogether. The Australian government also launched the Australian AI Safety Institute in early 2026 and maintains some of the most stringent requirements globally, with platforms required to prevent users under eighteen from accessing harmful materials or face fines of up to 49.5 million Australian dollars.

The contrast with the United States is stark. Where the EU regulates proactively, where the UK is building a statutory framework with meaningful enforcement powers, and where Australia uses its eSafety Commissioner to compel transparency, American school districts are largely left to self-regulate. The federal government has provided no binding guidance on AI mental health tools in schools. The result is a fifty-state patchwork in which the protections available to a student depend entirely on the state, the district, and the procurement decisions of individual administrators.

What Accountability Should Look Like

The current situation is untenable. Schools have a genuine need to support student mental health. AI tools offer genuine capabilities. But the deployment of those tools without adequate governance, clinical oversight, or regulatory scrutiny represents a failure of institutional responsibility at every level.

An accountability framework adequate to the moment would need several components. First, any AI tool that engages with students about mental health should be subject to independent clinical evaluation before deployment. This does not mean self-reported clinical trials funded by the vendor. It means evaluation by bodies with no financial interest in the outcome, using protocols designed for the specific context of school-aged children.

Second, parental consent should be meaningful, informed, and opt-in. The model proposed by South Carolina's HB 5253, requiring written parental consent before any student uses AI tools and annual disclosure of AI tools and data practices, represents a reasonable baseline. Parents cannot exercise judgement about tools they do not know exist.

Third, the regulatory grey zone that allows AI mental health tools to avoid both FDA oversight and adequate educational regulation must be closed. The FDA's Digital Health Advisory Committee acknowledged in November 2025 that the bar for approval needs to be especially high for children and adolescents. Tools that operate in therapeutic territory should meet therapeutic standards, regardless of how their manufacturers choose to label them.

Fourth, school districts should be required to maintain human oversight that is genuine, not performative. Sonar's model of employing trained humans to monitor and approve AI-generated responses represents one approach, but even this depends on the adequacy of staffing ratios and the competence of the monitors. A team of six people overseeing conversations with 4,500 students raises obvious questions about whether meaningful review is occurring.

Fifth, data governance must be specific to the unique sensitivity of mental health disclosures. Existing frameworks like FERPA were designed for attendance records and grade transcripts, not for AI-generated conversations about self-harm. Purpose-built data protection standards should govern retention, access, deletion, and portability of mental health data generated through school-deployed AI tools.

Sixth, there must be mandatory adverse event reporting. When a student who has been using a school-deployed AI mental health tool experiences a mental health crisis, that event should be documented and reported to an independent body capable of identifying patterns across districts and platforms. Currently, there is no such reporting requirement and no such body.

Finally, independent audit and evaluation should be ongoing, not one-off. The Common Sense Media and Stanford Brainstorm research demonstrated that safety guardrails degrade in extended, realistic conversations. A tool that passes an initial assessment may fail in the field. Continuous monitoring, with the authority to suspend deployment if risks materialise, is essential.

The Experiment Nobody Voted For

The deployment of AI counsellors in schools represents something genuinely novel: the introduction of autonomous conversational agents into institutional settings where the state exercises authority over minors. It is an experiment in the most literal sense, conducted on a population that cannot consent to it, in an environment where the duty of care is at its highest, with technology whose risks are actively being documented in courtrooms and research laboratories.

The people running this experiment are not villains. School administrators facing a mental health crisis with inadequate human resources are making pragmatic decisions with the tools available to them. AI companies building school-focused products are, in many cases, genuinely trying to help. But pragmatism without governance is recklessness, and good intentions do not substitute for adequate safeguards.

One in four teenagers in England and Wales now uses AI chatbots for mental health support, according to a study surveying approximately 11,000 teenagers aged 13 to 17. In the United States, approximately 5.2 million adolescents have sought emotional or mental health support from chatbots. Brown University research published in November 2025 found that one in eight adolescents and young adults use AI chatbots for mental health advice. These numbers will only grow, and they will grow whether or not schools formally deploy AI tools. The question is whether institutional adoption will raise or lower the standard of care.

Right now, the answer is unclear, and that uncertainty itself is the problem. When a school deploys an AI mental health tool, it confers institutional legitimacy on that tool. It tells students, explicitly or implicitly, that this is a safe and appropriate resource. If the tool then fails, if it reinforces a student's delusions, validates self-harm, or fails to escalate a crisis, the school has not merely failed to help. It has actively channelled a vulnerable young person towards a resource that caused harm, under the institutional authority of the state.

The lawsuits against Character.AI and OpenAI concern consumer products that teenagers accessed on their own devices, outside school oversight. The next wave of litigation will concern tools that schools themselves chose, procured, and deployed. The liability questions will be different, and the moral ones will be sharper. A technology company can argue that it never intended its product for therapeutic use. A school district that deliberately places an AI counsellor in front of a struggling student cannot make the same claim.

Twenty-five states are considering AI-in-education legislation. The EU AI Act is entering force. The UK is updating its safeguarding guidance. Australia is issuing transparency notices. These are steps in the right direction. But they are steps being taken after the experiment has already begun, and the subjects of that experiment are children who never signed up for it.

The counselling gap in schools is real and urgent. The desire to fill it is understandable. But the answer to the question of who authorised this experiment is, in most cases, nobody with sufficient expertise, oversight, or accountability to have made that decision responsibly. Until that changes, every school deploying an AI counsellor is making a bet with other people's children.

References

1. American School Counselor Association, “School Counselor Roles and Ratios,” schoolcounselor.org, 2024-2025 data.
2. TechCrunch, “This mental health chatbot aims to fill the counseling gap at understaffed schools,” 23 February 2025.
3. Maro, “Mental Health Screening for Schools,” meetmaro.com, accessed April 2026.
4. Centre on Reinventing Public Education, “Districts and AI: Early Adopters Focus More on Students in 2025-26,” crpe.org, 2025.
5. The Boston Globe, “AI chat bots are the new college counselors,” 25 March 2026.
6. CNN, “This mom believes Character.AI is responsible for her son's suicide,” 30 October 2024.
7. CNN, “Character.AI and Google agree to settle lawsuits over teen mental health harms and suicides,” 7 January 2026.
8. CNN, “Parents of 16-year-old Adam Raine sue OpenAI, claiming ChatGPT advised on his suicide,” 26 August 2025.
9. US Senate Judiciary Committee, “Written Testimony of Matthew Raine,” 16 September 2025.
10. TechCrunch, “OpenAI says over a million people talk to ChatGPT about suicide weekly,” 27 October 2025.
11. Common Sense Media, “Common Sense Media Finds Major AI Chatbots Unsafe for Teen Mental Health Support,” 20 November 2025.
12. RAND Corporation, “Teens Are Using Chatbots as Therapists. That's Alarming,” September 2025.
13. American Academy of Pediatrics, “Experts discuss potential benefits, harms, safeguards of using AI chatbots for mental health,” AAP News, 2025.
14. FutureEd, “Legislative Tracker: 2026 State AI in Education Bills,” future-ed.org, updated March 2026.
15. South Carolina Legislature, “2025-2026 Bill 5253: AI in Education,” scstatehouse.gov.
16. National Education Association, “Sample School Board Policy on AI Issues,” nea.org.
17. FDA Digital Health Advisory Committee, meeting on generative AI in digital mental health devices, 6 November 2025.
18. European Parliament, “Artificial Intelligence Act,” including Annex III on High-Risk AI Systems and provisions on children's vulnerability.
19. CNBC, “AI chatbot firms face stricter regulation in online safety laws protecting children in the UK,” 16 February 2026.
20. UK Department for Education, “Keeping Children Safe in Education 2026: Proposed Key Changes,” consultation document.
21. Australia eSafety Commissioner, “eSafety requires providers of AI companion chatbots to explain how they are keeping Aussie kids safe,” October 2025.
22. EdSource, “AI chatbots provide mental health support to 1 in 4 teenagers, study finds,” 2025.
23. Brown University School of Public Health, “One in eight adolescents and young adults use AI chatbots for mental health advice,” 18 November 2025.
24. RAND Corporation, “More Districts Are Training Teachers on Artificial Intelligence: Findings from the American School District Panel,” 2025.
25. Securly, “Student Wellness Monitoring Solution: Securly Aware,” securly.com, accessed April 2026.

---

Tim Green UK-based Systems Theorist & Independent Technology Writer

Tim explores the intersections of artificial intelligence, decentralised cognition, and posthuman ethics. His work, published at smarterarticles.co.uk, challenges dominant narratives of technological progress while proposing interdisciplinary frameworks for collective intelligence and digital stewardship.

His writing has been featured on Ground News and shared by independent researchers across both academic and technological communities.

ORCID: 0009-0002-0156-9795 Email: tim@smarterarticles.co.uk

Discuss...