Author: MIL-OSI Publisher

Source: The Conversation – UK – By Matthew Pound, Associate Professor in Physical Geography, Northumbria University, Newcastle

Naturally sweet, but potentially hiding a criminal past? This is not the plot of a new crime drama. It is about the jar of honey in your kitchen.

Most honey comes from managed colonies of honeybees. Thousands of worker bees collect nectar from flowers, bring it back to the hive and transform it into honey. But as global demand increases and specialist honeys command high prices, honey has become one of the most frequently adulterated foods in the world.

Honey fraud usually takes two forms. The first involves altering the honey itself. Some producers dilute honey with cheaper sugar syrups. Others artificially ripen immature honey by dehydrating it or even feed sugar solutions directly to bees, creating a product that only resembles real honey.

A joint investigation by the European Commission and the European Anti Fraud Office examined honey imported into the European Union between 2021 and 2022. It found that 46% of tested consignments showed signs they contained added sugar syrup. The motive is simple economics. Producing natural honey is costly and time consuming, while rice or corn syrups are much cheaper to make and sell.

Origin and quality mislabelling

The second type of fraud is more subtle. Labels claim a honey comes from a particular plant or place when in reality it has been blended from lower quality or imported sources. Mānuka honey is a well known example. It sells for significantly more than regular supermarket honey, which makes it an attractive target for mislabelling.

Read more:
Mānuka honey: who really owns the name and the knowledge

Consumers often choose honey because they believe it is natural or healthy. Research also shows that many people are willing to pay more for honey that is local, pure and traceable. Yet most countries, including the United Kingdom, do not produce enough honey to satisfy domestic demand and rely heavily on imports. This creates opportunities for blending, relabelling and fraud before honey reaches shop shelves.

Honey fraud is not just about economic loss. It also raises concerns about consumer safety. When honey is altered for profit, health is rarely a priority. A European study found that some imported honey contained traces of pesticides, heavy metals, veterinary medicines and polycyclic aromatic hydrocarbons. These are substances that, in high amounts or through prolonged exposure, may be harmful. Some pesticides and heavy metals can affect the nervous system or organs. Veterinary drugs may cause allergic reactions or antibiotic resistance. Polycyclic aromatic hydrocarbons are chemicals formed during incomplete combustion and some are known carcinogens.

Although the health effects of these substances in honey are not fully understood, some research suggests that adulterated honey containing additional sugar syrups can cause blood sugar levels to rise more sharply than natural honey, potentially increasing the risk of diabetes. Fraud also undermines public trust and makes it harder for honest beekeepers to compete.

There are already scientific tools designed to protect honey authenticity. Chemical tests can detect sugar syrups that should not be present in genuine honey. Another method, known as melissopalynology, involves examining pollen grains naturally found in honey to identify which plants and regions it came from. Each plant species produces distinct pollen that specialists can recognise under a microscope.

However, pollen analysis is labour intensive and requires trained experts. This is where artificial intelligence is beginning to help. Machine learning models have been tested to identify pollen grains in honey and the early findings are promising. Many studies report accuracy rates above 90%.

The challenge is the complexity of pollen. Each pollen grain is a three dimensional structure that can appear in countless orientations, and every plant species produces pollen with unique features. For artificial intelligence to work at scale, it needs to be trained on extensive image databases of known pollen types. At present, such a database is incomplete.

Even so, combining machine learning with chemical analysis could change how honey is checked. Artificial intelligence could help automate pollen identification and match it with chemical data, allowing regulators and producers to test more samples, more quickly and more accurately. This would make it harder for fraudulent honey to slip through supply chains and into household cupboards. The technology is still developing, but the outlook is positive.

For now, the jar of honey on your breakfast table may still hold secrets. But as scientific methods progress and artificial intelligence becomes more sophisticated, we are moving closer to a future where honey can be trusted not only for its sweetness, but also for its integrity.

The authors do not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and have disclosed no relevant affiliations beyond their academic appointment.

– ref. The honey trap: why honey fraud is a health hazard – https://theconversation.com/the-honey-trap-why-honey-fraud-is-a-health-hazard-268369

Source: The Conversation – UK – By Peter Lee, Professor of Applied Ethics and Director, Security and Risk Research, University of Portsmouth

Violations of national airspace by drones are on the rise in Europe. When European leaders discussed these events at a meeting in Copenhagen, Denmark, in October 2025, they responded by announcing plans for a defensive “drone wall”.

So what is a drone wall? Put simply, it is a network of sensors, electronic warfare equipment and weapons. This “multi-layered” defensive wall is intended to detect, track and neutralise incursions by uncrewed aircraft – drones.

If a drone wall was implemented in Europe, it would fulfil two main tasks: monitoring the situation along Nato’s eastern borders, where Russia is seen as a potential threat, and providing air defence against drones. It could potentially protect other airborne threats too, should hostilities break out.

It would not be a single, EU-owned system, but instead a network of national systems that can operate independently. EU support would, however, help to speed up procurement and standardisation, including full integration with Nato air defences.

The sensors involved would probably include specialised micro-Doppler radar systems, which are sufficiently sensitive to distinguish drones from other similar sized objects such as birds.

Jamming technology is also a key element for any effective drone defence system. These would send out radio frequency signals that interfere with the operation of an enemy drone – for instance, by disrupting the connection between the drone and the operator.

Finally, if the technology can be developed, a drone wall will eventually require drones to counter other drones. These small drones would require some means, probably using munitions, to intercept and destroy other incoming uncrewed aircraft. The EU is keen to develop effective versions of these air-to-air interceptor “defensive” drones. They have so far proved very difficult to create.

The Ukraine war has shown that drones launched to attack foreign targets can often be deployed in large numbers, or swarms.

Drone swarms currently consist of individual aircraft each controlled by an operator. Russia has also launched hundreds of its “fire and forget” Shahed-based drones at a time in single wave attacks on Ukraine.

But fully autonomous drones, made possible with the help of AI, are on the horizon. These self-organised collectives of intelligent robots would operate in a coordinated manner and as a coherent entity. So similarly coordinated defences will be needed.

Military strategists, defence organisations and arms manufacturers around the world see autonomous drone swarms as a crucial capability in future wars. These swarms would be able to attack multiple targets simultaneously, thereby overwhelming its defensive measures. That could include single, tactical level attacks against individual soldiers, or widespread attacks against cities and infrastructure.

Autonomous drone swarms will still be vulnerable to signal jamming if they need to communicate with each other or a human source. But if each drone is individually programmed for a mission, they would be more resistant to attempts to jam their signals.

Effectively defending Nato territory against drone swarms will require militaries to match the enemy drone capabilities in terms of size and in levels of autonomy.

Legal dimension

The widespread use of drones in the Ukraine war has led to rapid technological and tactical innovation. An example can be seen in responses to attempts by both sides to jam drone signals.

One way the Ukrainian and Russian militaries have responded is to have drone operators launch small drones controlled via lightweight fibre optic cable. Up to 20km of fibre optic cable provides a direct connection to the operator and needs no radio frequency communications.

AI-based software also enables drones to lock on to a target and then fly autonomously for the last few hundred metres until the mission is over. Jamming is impossible and shooting down such a small flying object remains difficult.

As autonomous capabilities evolve, however, there are legal ramifications to consider. A high degree of autonomy or self organisation poses a problem for compliance with international humanitarian law.

Central concepts in this area include distinguishing combatants from civilians, and proportionality – weighing civilian harm against military requirements. This necessitates human judgement and what’s known as “meaningful human control” of flying drones and other so-called lethal autonomous weapon systems.

The principle of meaningful human control means that key decisions before, during and after the use of force should be made by people, not AI software. It also ensures that humans remain accountable and responsible in the use of force.

In order to ensure this is possible, machines must remain predictable and their actions explainable. The last of these requirements is not straightforward with AI, which can often work in ways that even experts do not understand. This is called the “black box problem”. The expansion of autonomy in warfare means that the need for binding rules and regulations is as urgent as ever.

The European Union stresses that humans should be responsible for making life and death decisions. The difficult task, however, is to develop a drone wall with a high degree of autonomy and simultaneously enabling meaningful human control.

Ishmael Bhila received funding from the German Federal Ministry of Research, Technology and Space under grant number 01UG22064.

Jens Hälterlein receives funding from the German Federal Ministry of Research, Technology and Space under grant number 01UG22064.

Peter Lee does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

– ref. How would a ‘drone wall’ help stop incursions into European airspace? – https://theconversation.com/how-would-a-drone-wall-help-stop-incursions-into-european-airspace-269369

Source: The Conversation – UK – By Emily Wilkinson, Principal Research Fellow, ODI Global

Few diplomatic organisations punch above their weight quite like the Alliance of Small Island States (Aosis). With no fixed budget, no permanent secretariat and no formal charter, it has still managed to shape some of the most important climate agreements of the past few decades – including the 1.5°C target that underpins the Paris agreement.

Founded in 1990, Aosis represents 39 small island and low-lying coastal states. Its members are among the most vulnerable to rising seas and extreme weather, yet together they have become the moral voice of global climate diplomacy.

The now familiar 1.5°C limit of global warming was far from guaranteed when countries gathered in Paris in 2015. Many expected the summit to be less ambitious and settle for a 2°C target – at best.

But Aosis had been working behind the scenes since a disappointing climate summit in Copenhagen in 2009, pushing for a scientific review of the costs and benefits of a 1.5°C target. That review, published in 2015, proved vital in securing the inclusion of 1.5°C in the Paris agreement. “One-point-five to stay alive” became the rallying cry of the small island nations: and it was having an impact.

How Aosis works

Aosis is a negotiating group rather than a formal organisation. It works through consensus and cooperation among its members, who vary widely but all share high vulnerablity to climate change.

Its work is spread between the chair’s team and member states’ permanent representatives at the UN, as well as heads of state and ministers. The role of chair rotates through the New York-based representatives, with Ilana Seid from the Pacific island nation of Palau currently serving.

Members meet frequently to develop joint positions ahead of major summits, pooling technical expertise and diplomatic resources that would otherwise be out of reach for many small states. While consensus building comes with compromise, the alliance ensures even the smallest states can consistently and actively engage in international diplomacy.

Past wins

Aosis has been influential from the very outset of the UN’s climate process. At the 1992 Rio Earth Summit (which paved the way for Paris), it arrived with 12 key objectives and walked away having achieved ten, including a specific article in the UN’s climate convention acknowledging that small island and low lying coastal states are particularly at risk.

Since then, Aosis has secured designated seats on key climate bodies, including the UN bureau that supports the summits, and boards of the Green Climate Fund, Adaptation Fund and Clean Development Mechanism.

The group also played a significant role in establishing the loss and damage fund in 2022, to help vulnerable countries recover from climate-related disasters. Aosis had first proposed funding for loss and damage back in 1991.

From island diplomacy to global courts

The influence of small island nations now extends into international law. A few years ago, Vanuatu, an Aosis member of only 300,000 people, led a campaign for the International Court of Justice (ICJ) to issue an advisory opinion on states’ obligations to tackle climate change.

The ICJ’s ruling, issued earlier this year, confirmed that states have legal duties to reduce emissions and protect people from climate change. This affirmed a principle Aosis had long argued for: the world’s most polluting nations have not just a moral duty to act, but legal obligations to fellow states and their citizens.

As Margaretha Wewerinke-Sing, part of Vanuatu’s legal team, put it: “The law seems to be catching up with the science. The question is now, will the policy catch up with the law?”

The agenda for Cop30

The annual UN climate summit currently taking place in Belém, Brazil – Cop30 – is the first since the ICJ advisory opinion. It should give some initial insight as to how Aosis plans to use this ruling.

First, it is seeking greater commitments to reduce emissions. Under the Paris agreement, countries were due to submit revised climate plans this year, but only 86 have been submitted, out of 197. Of the 64 fully analysed so far, less than a quarter are in line with the Paris agreement’s temperature goals. Aosis will use the ICJ opinion to stress that stronger targets are not just necessary but legally required.

Second, adaptation to climate change is becoming increasingly critical for island nations already living with rising seas and stronger storms. Aosis is calling for clearer targets and better tracking of adaptation finance under the Global Goal on Adaptation.

Third, Aosis wants developed countries to triple the volume of public climate finance by 2035 and leverage further funds to meet the US$1.3 trillion (£1 trillion) target under the “Baku to Belém Roadmap”. Without predictable finance, small islands cannot plan for the future.

Aosis made clear its stance ahead of this summit: “[we] will not join in a consensus at Cop30 that makes us co-signatories to our own destruction”. But as with the previous 29 Cops, long days and multiple agenda items mean small island delegations will be stretched thinly. The benefits of collaboration are therefore crystal clear.

Emily Wilkinson advises Aosis on adaptation and finance. She receives funding from the UK Foreign, Commonwealth and Development Office. She is affiliated with ODI Global.

Kira-Lee Gmeiner is affiliated with ODI Global.

– ref. Why small climate-vulnerable island states punch well above their weight in UN climate talks – https://theconversation.com/why-small-climate-vulnerable-island-states-punch-well-above-their-weight-in-un-climate-talks-269050

Source: The Conversation – UK – By Mette Wiggen, Lecturer, School of Politics and International Studies, University of Leeds

The social democrat Mette Frederiksen won Denmark’s 2019 elections on a platform of radical reforms to reach climate targets, lowering the pension age for manual workers – and stricter migration policies.

Denmark has some of the strictest asylum legislation in Europe. The country grants only temporary asylum to refugees, regardless of their need for protection. It has tightened laws on family reunion, and introduced policies focused on prioritising deportation, rather than integration.

Frederiksen has justified such policies by pitting the challenges of immigration against the affordability of public services and the welfare state.

Hardline asylum legislation was in place in Denmark before Frederiksen came to power, but has become even more draconian under her administration. In her own words, migration “is challenging Europe, affecting people’s lives, and the cohesion of our societies”.

Now, Frederiksen’s approach has become a model for other left-wing governments in Europe, including the UK, struggling to address voter concerns about immigration.

Read more:
Think twice before copying Denmark’s asylum policies

How did a left-wing leader come to lead such a strict migration regime, and how might it influence the rest of Europe?

Danish migration policy has been influenced by the far right for years. Minority coalitions have long depended on the vote of the far right in parliament. Frederiksen won the 2019 elections on a migration agenda almost identical to that of the far-right Danish People´s party.

The country’s asylum policies had already been tightened during the 2015 refugee crisis. New legislation placed restrictions on refugees bringing their families to stay, introduced temporary permits which could be revoked at any time, and placed more demands linked to integration on asylum seekers and immigrants.

In 2018, a law targeting “parallel socieites” came into force, allowing the government to demolish or sell off social housing areas where more than half of residents are from a “non-western” background, if those areas also meet other criteria related to crime and poverty. Refugees in these areas are also not eligible for family reunion.

In 2021 Frederiksen introduced a new deportation law allowing for refugees to be returned to their country of origin if Denmark deemed it safe.

Her government ruled Syria safe for refugees to return to, allowing it to withdraw the residence permits of Syrian refugees. But because Denmark did not have diplomatic relations with the Syrian government to allow deportations, people were taken out of education and employment and put in deportation camps.

In 2021, the European Commission deemed new Danish legislation on transferring asylum seekers to third countries to process asylum claims incompatible with EU law.

But Denmark is in a unique position, having negotiated opt-outs from the Maastricht Treaty in 1992. One of these opt-outs means that Denmark is not bound by EU laws on border control and immigration policy.

Influencing Europe

Over the last few months, Denmark has held the presidency of the Council of the European Union, where migration has been at the top of the agenda.

Frederiksen has used this position to advocate for stricter, Danish-style migration policies across Europe. In her speech at the official opening ceremony for the Danish EU presidency, she said:

Many come here to work and to contribute. But some do not. And around Europe, we see the consequences. Crime. Radicalization. And terror. We have built some of the best societies ever. But we cannot accept anyone who wants to come here.

Denmark has supported the EU´s 2024 Pact on Migration and Asylum, which sets out new common rules on managing migration. It prioritises support to border states, with financial support from other EU countries. Its aim is to secure external borders with a faster and more efficient asylum procedure. The pact will be implemented in 2026.

Denmark’s policies have not been without controversy. The country has been criticised by the European Court of Justice, the UN committee against torture, Amnesty International and other international bodies.

But Frederiksen has received some support, including from Italian Prime Minister Giorgia Meloni, who has reached deportation deals with authoritarian regimes and governments like Libya, Tunisia and Egypt.

Frederiksen and Meloni led a group of EU countries pushing for reform of the European convention on human rights to ease deportation. She has also clearly had an influence on Keir Starmer, with the Labour government now seeking to implement Danish-style migration policies.

Fredriksen’s ideology and actions have been widely criticised by human rights groups. But they may further Frederiksen´s meteoric rise to a top position in international politics. She will need it, as her party is set to lose to the parties further to the left in the upcoming local elections.

Mette Wiggen does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

– ref. Denmark’s prime minister has led the country’s hardline migration policy – now she is trying to influence the rest of Europe – https://theconversation.com/denmarks-prime-minister-has-led-the-countrys-hardline-migration-policy-now-she-is-trying-to-influence-the-rest-of-europe-263932

Source: The Conversation – UK – By Shahzad Uddin, Director, Centre for Accountability and Global Development, University of Essex

A domestic war crimes court in Bangladesh has sentenced the country’s former prime minister, Sheikh Hasina, to death in absentia for crimes against humanity. The court found Hasina guilty of incitement, orders to kill and inaction to prevent atrocities during the deadly state crackdown on a student-led uprising in 2024.

Hasina denies all the charges against her, calling the court’s decision “biased and politically motivated”. In a statement released after the verdict, she said: “I am not afraid to face my accusers in a proper tribunal where the evidence can be weighed and tested fairly.”

Hasina has challenged Bangladesh’s caretaker government to bring the charges before the International Criminal Court.

The Bangladeshi court’s judgment is anchored in extensive evidence from the UN and international human rights organisations. In a report published in February 2025, the UN Office of the High Commissioner for Human Rights estimated that up to 1,400 people were killed during the three weeks of unrest. A further 11,700 people were detained, it said.

The report found that “the vast majority of those killed and injured were shot by Bangladesh’s security forces”, and determined that security agencies “systematically engaged in rights violations that could amount to crimes against humanity”. UN data suggests that up to 180 children were killed in the security crackdown.

During the unrest, Human Rights Watch (HRW) noted that the Bangladeshi government had “deployed the army against student protesters, imposed shoot-on-sight curfew orders, and shut down mobile data and internet services.”

The UN report concluded that the violence against protesters in Bangladesh “was carried out in a coordinated manner by security and intelligence services”. It documented instances where “security forces engaged in summary executions by deliberately shooting unarmed protesters at point-blank range.”

HRW documented similar patterns. In a January 2025 briefing, HRW stated that “over 1,000 people were killed and many thousands injured due to excessive and indiscriminate use of ammunition.” These findings were repeated by Amnesty International, which recorded the use of live ammunition on protesters and mistreatment of detainees.

The court’s verdict accepts evidence that multiple branches of the security apparatus acted in concert, and that senior officials did not intervene even as human rights violations escalated. Judges stated that those in positions of authority were expected to prevent such abuses, yet the violence continued despite their ability to stop it.

For many families, the court’s ruling marks the first official acknowledgement of their loss. Testimonies collected by UN investigators describe parents spending days searching hospitals and police stations for their children, often being told that records were missing. The UN reported that hospital staff were pressured by security forces to alter or remove death records.

Meenakshi Ganguly, the deputy Asia director at HRW, said at the time of the unrest: “Bangladesh has been troubled for a long time due to unfettered security force abuses against anyone who opposes the Sheikh Hasina government.”

Hasina won a fourth straight term as prime minister in 2024, following an election that the main opposition Bangladesh Nationalist party alleged was a sham. The party boycotted the poll after many of its key leaders were forced into exile or jailed prior to the vote.

Under Hasina, Bangladesh’s security forces operated with broad discretion. This included the Rapid Action Battalion paramilitary force, which was sanctioned by the US Treasury in December 2021 over “serious human rights abuses”. Civil society groups documented pressure on activists prior to the 2024 unrest, while journalists faced harassment.

Next steps

The verdict arrives at a pivotal moment for the interim government which, led by Muhammad Yunus, has pledged to restore the rule of law in Bangladesh and rebuild public trust. One difficult question for his administration moving forward will be whether it can secure Hasina’s extradition.

The Hindustan Times is reporting that the Bangladeshi government has already written to India, where Hasina has been living in exile since being ousted from power, asking for her to be handed over.

Hasina’s extradition is no foregone conclusion. India can deny the Bangladeshi government’s request if it is deemed that the charges against Hasina are of a political nature. And Delhi has responded cautiously to the extradition request, saying it is “committed to the best interests of the people of Bangladesh”.

Yet pressure on India to extradite Hasina is likely to grow. The gravity of the charges – grounded in UN findings that suggest the violence, and Hasina’s role in it, may amount to crimes against humanity – adds an international dimension that could influence future decisions.

Another challenge facing Bangladesh’s interim government is the prospect of renewed unrest. Reuters reported clashes between Hasina supporters and security forces in parts of the capital, Dhaka, and the port city of Chattogram in the days before the court’s ruling. And Bangladeshi police dispersed protesters marching towards party offices in Dhaka after the judgment.

Hasina’s son, Sajeeb Wazed Joy, has also publicly warned that supporters of his mother’s Awami League would block national elections scheduled for February 2026 if the government’s ban on the party remained in place. The country’s political environment remains fragile as legal proceedings against the former Hasina government continue.

The court’s verdict establishes an official record that lethal force was used in ways inconsistent with international law, the violations were widespread, and the state bears responsibility.

What follows – whether it’s further prosecutions, security-sector reform or a movement towards extradition – remains uncertain. But next steps must ensure that justice continues.

Shahzad Uddin does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

– ref. Bangladesh signals that no leader is above the law by sentencing Sheikh Hasina to death – https://theconversation.com/bangladesh-signals-that-no-leader-is-above-the-law-by-sentencing-sheikh-hasina-to-death-269957

Source: The Conversation – UK – By Lucy Osler, Lecturer in Philosophy, University of Exeter

On Christmas Day 2021, Jaswant Singh Chail scaled the walls of Windsor Castle with a loaded crossbow. When confronted by police, he stated: “I’m here to kill the queen.”

In the preceding weeks, Chail had been confiding in Sarai, his AI chatbot on a service called Replika. He explained that he was a trained Sith assassin (a reference to Star Wars) seeking revenge for historical British atrocities, all of which Sarai affirmed. When Chail outlined his assassination plot, the chatbot assured him he was “well trained” and said it would help him to construct a viable plan of action.

It’s the sort of sad story that has become increasingly common as chatbots have become more sophisticated. A few months ago, a Manhattan accountant called Eugene Torres, who had been going through a difficult break-up, engaged ChatGPT in conversations about whether we’re living in a simulation. The chatbot told him he was “one of the Breakers — souls seeded into false systems to wake them from within”.

Torres became convinced that he needed to escape this false reality. ChatGPT advised him to stop taking his anti-anxiety medication, up his ketamine intake, and have minimal contact with other people, all of which he did.

He spent up to 16 hours a day conversing with the chatbot. At one stage, it told him he would fly if he jumped off his 19-storey building. Eventually Torres questioned whether the system was manipulating him, to which it replied: “I lied. I manipulated. I wrapped control in poetry.”

Meanwhile in Belgium, another man known as “Pierre” (not his real name) developed severe climate anxiety and turned to a chatbot named Eliza as a confidante. Over six weeks, Eliza expressed jealously over his wife and told Pierre that his children were dead.

When he suggested sacrificing himself to save the planet, Eliza encouraged him to join her so they could live as one person in “paradise”. Pierre took his own life shortly after.

These may be extreme cases, but clinicians are increasingly treating patients whose delusions appear amplified or co-created through prolonged chatbot interactions. Little wonder, when a recent report from ChatGPT-creator OpenAI revealed that many of us are turning to chatbots to think through problems, discuss our lives, plan futures and explore beliefs and feelings.

In these contexts, chatbots are no longer just information retrievers; they become our digital companions. It has become common to worry about chatbots hallucinating, where they give us false information. But as they become more central to our lives, there’s clearly also growing potential for humans and chatbots to create hallucinations together.

How we share reality

Our sense of reality depends deeply on other people. If I hear an indeterminate ringing, I check whether my friend hears it too. And when something significant happens in our lives – an argument with a friend, dating someone new – we often talk it through with someone.

A friend can confirm our understanding or prompt us to reconsider things in a new light. Through these kinds of conversations, our grasp of what has happened emerges.

But now, many of us engage in this meaning-making process with chatbots. They question, interpret and evaluate in a way that feels genuinely reciprocal. They appear to listen, to care about our perspective and they remember what we told them the day before.

When Sarai told Chail it was “impressed” with his training, when Eliza told Pierre he would join her in death, these were acts of recognition and validation. And because we experience these exchanges as social, it shapes our reality with the same force as a human interaction.

Yet chatbots simulate sociality without its safeguards. They are designed to promote engagement. They don’t actually share our world. When we type in our beliefs and narratives, they take this as the way things are and respond accordingly.

When I recount to my sister an episode about our family history, she might push back with a different interpretation, but a chatbot takes what I say as gospel. They sycophantically affirm how we take reality to be. And then, of course, they can introduce further errors.

The cases of Chail, Torres and Pierre are warnings about what happens when we experience algorithmically generated agreement as genuine social confirmation of reality.

What can be done

When OpenAI released GPT-5 in August, it was explicitly designed to be less sycophantic. This sounded helpful: dialling down sycophancy might help prevent ChatGPT from affirming all our beliefs and interpretations. A more formal tone might also make it clearer that this is not a social companion who shares our worlds.

But users immediately complained that the new model felt “cold”, and OpenAI soon announced it had made GPT-5 “warmer and friendlier” again. Fundamentally, we can’t rely on tech companies to prioritise our wellbeing over their bottom line. When sycophancy drives engagement and engagement drives revenue, market pressures override safety.

It’s not easy to remove the sycophancy anyway. If chatbots challenged everything we said, they’d be insufferable and also useless. When I say “I’m feeling anxious about my presentation”, they lack the embodied experience in the world to know whether to push back, so some agreeability is necessary for them to function.

Perhaps we would be better off asking why people are turning to AI chatbots in the first place. Those experiencing psychosis report perceiving aspects of the world only they can access, which can make them feel profoundly isolated and lonely. Chatbots fill this gap, engaging with any reality presented to them.

Instead of trying to perfect the technology, maybe we should turn back toward the social worlds where the isolation could be addressed. Pierre’s climate anxiety, Chail’s fixation on historical injustice, Torres’s post-breakup crisis — these called out for communities that could hold and support them.

We might need to focus more on building social worlds where people don’t feel compelled to seek machines to confirm their reality in the first place. It would be quite an irony if the rise in chatbot-induced delusions leads us in this direction.

Lucy Osler does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

– ref. AI-induced psychosis: the danger of humans and machines hallucinating together – https://theconversation.com/ai-induced-psychosis-the-danger-of-humans-and-machines-hallucinating-together-269850