Category: Analysis

Source: The Conversation – Global Perspectives – By Daniel Thomas, Honorary Lecturer in Biological Sciences, University of Auckland, Waipapa Taumata Rau

Three million years ago, an extinct relative of todays’s great penguins – emperors and kings – lived in Aotearoa New Zealand.

We know this because our new study describes a spectacular fossilised skull of a great penguin found on the Taranaki coast.

Overall, it is 31% longer than the skull of an emperor penguin (Aptenodytes forsteri), which can be more than a metre tall and weigh upwards of 35 kilograms.

Compared to emperor penguins, however, the Taranaki great penguin had a much stronger and longer beak. It probably looked more like a king penguin (Aptenodytes patagonicus), only much bigger.

At the time, the world was warmer than today. But when the climate cooled, this penguin vanished.

We argue the cold wasn’t to blame because crested and little penguins in New Zealand weathered the same change and remained. Great penguins shifted south and today live in the frozen wastes of Antarctica. So what drove their ancient relative to extinction?

The sediments that now form beach-side cliffs in South Taranaki were deposited at a time when global temperatures were about 3°C above those of the pre-industrial era. Fossils from this period are transforming our understanding of how biodiversity might respond to rising temperatures.

For example, Aotearoa was home to box fish and monk seals, both of which are still (sub)tropical species today. In a strange contradiction, they coexisted with great penguins – now only found in much colder climates – in ancient New Zealand.

The northernmost breeding colonies of king penguins today are around latitude 46.1°S in the subantarctic Crozet Islands, where seawater temperatures reach 3-10°C. From there, it only gets colder towards the higher latitudes where emperor penguins live.

Three million years ago, Aotearoa’s great penguins extended as far north as 40.5°S, where South Taranaki was located then. They foraged in waters that were 20°C, much warmer than their relatives experience today.

This balmy existence ended with the Pleistocene ice ages around 2.58 million years ago. Ice extent and sea level shifted back and forward as temperatures fluctuated and ultimately ratcheted downwards. But why would such cooling eradicate giant penguins, which thrive under polar conditions today, from New Zealand?

Giant aerial predators

Fossil evidence for giant penguins in Aotearoa is limited and the exact reasons for their demise remain unclear. Even so, their sheer presence suggests they were less constrained by sea surface temperatures than previously thought. Another mechanism must be at play.

Up until about 500 years ago, Aotearoa was the hunting ground of the giant Haast’s eagle and the huge Forbes’ harrier. These were big raptors. They included large birds like moa in their diet. Their ancestors arrived from Australia inside the last three million years.

Based on what we see with living great penguins, the Taranaki great penguin almost certainly formed large exposed colonies along the coast. These could have been easy targets for a giant eagle or harrier hunting from the air.

By contrast, the smaller penguins still found in Aotearoa today have more cryptic breeding behaviour. They nest in burrows, natural crevices and dense vegetation, and tend to cross beaches at night, which may have helped them avoid aerial predators.

Predation on land is just one hypothesis, though, to help explain why these penguins became extinct in the region while others survived. Other possibilities include changes in the marine environment.

We know that reduced food availability can be devastating for penguins, but it is challenging to see why this would single out the great penguins.

Importantly, our study provides new insight into the habitat tolerances of great penguins. Both king and emperor penguins today can withstand temperatures up to 20°C higher than those they usually forage in.

Three million years ago, their relative experienced such warmth. As the world continues to warm, we need to remember that the geographic range of a species can change as circumstances change.

The marine ecosystem of Aotearoa will move into the habitable zone of many new species, making investigations of the last warm period more important than ever before.

---

We would like to acknowledge our research co-author Dan Ksepka from The Bruce Museum, Kerr Sharpe-Young for discovering the fossil, and Ngāti Ruanui and Ngāruahine for supporting the collection and research of fossils from their rohe.

---

Daniel Thomas has received funding from Massey University.

Alan Tennyson and Felix Georg Marx 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. Close relatives of emperor penguins lived in NZ some 3 million years ago. What caused their extinction? – https://theconversation.com/close-relatives-of-emperor-penguins-lived-in-nz-some-3-million-years-ago-what-caused-their-extinction-265585

Source: The Conversation – Global Perspectives – By Fred H. Lawson, Professor of Government Emeritus, Northeastern University

President Donald Trump chided the United Nations on Sept. 23, 2025, for failing to resolve dangerous international conflicts around the world. “All they seem to do,” he groused during his address to the General Assembly in New York, “is write a really strongly worded letter and then never follow that letter up. It’s empty words, and empty words don’t solve war.”

In contrast, Trump, by his own estimation, has ended a half-dozen serious international conflicts. Among them is the long-standing dispute over the Nile River waters that flared up when Ethiopia proposed to build a massive dam over the Blue Nile, threatening the water supply of Egypt and Sudan.

As an international relations scholar who has studied this dispute, however, I find it hard to see how Trump’s interventions have brought it any closer to a resolution. In fact, they most likely made things worse.

The source of the dispute

Nile River water is essential to agriculture and public sanitation in both Egypt and Sudan. Distribution of that water has been regulated by an agreement drawn up in 1959 guaranteeing Egypt and Sudan most of the output of the Nile River basin. A body of international law also enjoins upstream nations – such as Ethiopia – against manipulating the cross-border flow of rivers in ways that harm downstream countries.

Ethiopia nonetheless declared in 2011 that it had the right to exploit any water resources that originate inside its borders. It further insisted that constructing an electricity-generating dam across the Blue Nile would provide cheap power to both impoverished Ethiopians and neighboring countries.

Egypt and Sudan immediately protested on the grounds that the project would inflict severe damage on their populations and implored international organizations and external powers to intervene.

Ambiguous wording

Trump waded into the Nile dispute at the urging of Egypt. Cairo approached Washington to mediate in late October 2019, as Ethiopia was ramping up construction of the Grand Ethiopian Renaissance Dam that would restrict the flow of the Blue Nile. After speaking personally with Egypt President Abdel Fattah el-Sissi, Trump agreed to become directly involved.

He and then-Secretary of the Treasury Steven Mnuchin invited the foreign ministers of Egypt, Sudan and Ethiopia to Washington for talks.

With the opening session of the ensuing November 2019 meeting still ongoing, Trump tweeted: “The meeting went well and discussions will continue during the day!”

The talks, however, ended with no progress being made. Instead, all four parties, along with representatives of the World Bank, agreed to confer a dozen more times over the next three months to hash out technical matters.

At these subsequent meetings, Egypt and Ethiopia haggled over definitions and measurement criteria concerning the dam’s possible impact. Mnuchin and his staff largely stood aside, although they reportedly expressed sympathy for Addis Ababa’s insistence that matters concerning the Grand Ethiopian Renaissance Dam be kept separate from questions concerning water management in the Nile Basin as a whole.

More importantly, U.S. representatives allowed imprecise wording to slip into the
statement released at the close of the December 2019 meeting.

The statement mandated that “the implementation of these technical rules and guidelines for the filling and operation of the dam will be undertaken by Ethiopia, and may be adjusted by the three countries.”

Cairo interpreted this to mean that all regulations and procedures would be drawn up jointly; Addis Ababa believed that it enshrined Ethiopia’s right to make decisions entirely on its own.

Talks blow up

The final round of talks in January 2020 produced a prospective agreement that left Ethiopia free to start filling the huge reservoir behind the dam, and minimized the country’s obligations to assist Egypt and Sudan during drought periods.

Yet, Ethiopia dragged its feet in accepting the draft document, claiming that crucial points remained unsettled. Egypt and Sudan flatly refused to revisit problems they believed had already been addressed, but did accept an offer from U.S. Treasury officials to prepare a revised text.

In February 2020, the U.S. Treasury Department circulated an amended version, and Trump telephoned al-Sissi to express hope that “an agreement would be finalized soon.”

Ethiopia, feeling pressured by Egypt and the U.S. to endorse an incomplete text – and worried about the domestic political consequences of doing so – did not send a representative to Washington to accept the revision. Treasury officials then declared that a comprehensive settlement to the dispute had now been reached and publicly called on Addis Ababa to sign it.

Egypt’s foreign ministry issued a statement asserting that “President Trump affirmed the U.S. administration’s continued efforts” to reach an acceptable deal. But Ethiopia’s foreign minister described Washington’s abrupt declaration as “undiplomatic.”

Ethiopia stalled for another two months, then circulated a provisional agreement of its own, which Egypt and Sudan dismissed out of hand.

Washington responded with insinuations that it would withhold economic assistance from Ethiopia unless Addis Ababa signed the agreement the Treasury Department had drafted in February 2020. Meanwhile, construction on the dam proceeded, and in July Ethiopia blocked the flow of the Blue Nile to begin filling its huge reservoir.

In September 2020, then-Secretary of State Mike Pompeo followed through on the U.S. threat and suspended US$130 million of aid to Ethiopia. However, the suspension had no impact on negotiations.

Incensed at the continuing impasse, Trump in October 2020 remarked during a telephone call with Sudanese and Israeli diplomats that Egypt “will end up blowing up the dam.”

Ethiopian officials condemned the outburst. The prime minister’s office complained that “these threats and affronts to Ethiopian sovereignty are misguided, unproductive and clear violations of international law.” Ethiopia installed anti-aircraft batteries around the dam and declared the airspace above it a no-fly zone.

Negotiations collapsed, and they remained dormant for the next three years.

Assessing the impact of US mediation

Analysts tend to agree that Trump’s initial involvement in the Nile River dispute made a bad situation worse. His admiration for authoritarian leaders prompted him to accommodate Egypt’s al-Sissi and put U.S. credibility on the line.

Meanwhile, Trump’s contempt for the professionals in the U.S. State Department led him to sideline the diplomatic corps and entrust a complicated assignment to Mnuchin, a former financier and movie producer. And his lack of patience and blunt language disrupted the negotiations, alienating Egypt and Ethiopia alike.

Addis Ababa continues to insist that it “has no obligation to request permission from anyone to fill the Renaissance Dam.” In September 2025, Prime Minister Abiy Ahmed declared that work on the dam had finished and boasted of two other dams on the Blue Nile nearing completion.

Meanwhile, Egypt has opened a large naval base on the Red Sea coast and attached
its most advanced warships to a newly created Red Sea squadron. In July 2025, Egypt Foreign Minister Badr Abdelatty issued a veiled threat to use military means to settle the Nile dispute.

Despite rising tensions, and his first administration’s failure to make progress on the issue, Trump continues to point to his 2019-2020 mediation as a success.

In July 2025, he told NATO Secretary General Mark Rutte that a resolution lay just around the corner. President al-Sissi once again applauded the president’s involvement and voiced hope that it would produce a “just agreement.”

Yet there is little indication that the current Trump administration is in any better position to solve the Nile River dispute than was the previous one. Since Trump’s second inauguration, experienced State Department officials have been fired or resigned, leaving sensitive diplomatic missions in the hands of private businesspeople with personal ties to the president, rather than diplomats skilled in the art of negotiating intractable disputes among sovereign states.

This development seems unlikely to budge Trump from the conviction that he has already solved the conflict over the Nile River waters – and can somehow do it again.

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

– ref. Trump’s dip into the Nile waters dispute didn’t settle the conflict – in fact, it may have caused more ripples – https://theconversation.com/trumps-dip-into-the-nile-waters-dispute-didnt-settle-the-conflict-in-fact-it-may-have-caused-more-ripples-263767

Source: The Conversation – Global Perspectives – By Will Brehm, Associate Professor of Comparative and International Education, University of Canberra

In late July, during a visit to the National Gallery of Australia, three Buddhist bodhisattva statues caught my attention.

All three were created in the ancient Champa Kingdom that flourished from the 2nd to 19th centuries across present-day Vietnam, Cambodia and Laos. They were purchased by the National Gallery (NGA) in 2011, before being “repatriated” to the Kingdom of Cambodia in 2023 (and displayed in the NGA on loan).

But the Champa Kingdom bore little resemblance to Cambodia’s current borders. What does repatriation mean when the political geography of a place has entirely transformed?

As my research has shown, museums, schools and state institutions can help sanction certain versions of history, while marginalising others. The quiet presence of the bodhisattvas in a museum case embodies much larger questions about cultural heritage, political legitimacy, and who gets to define historical “truth”.

Decades of marginalisation

The decision to return the Cham artefacts to Cambodia, and to exclude Vietnam and Laos, highlights how contemporary politics shape our understanding of cultural heritage.

The Cham people are an ethnic minority in Cambodia, Vietnam, and Laos. In Cambodia, they have been marginalised by the ruling government’s Khmer ethno-nationalist vision of the country.

Although most Cham people today are Muslim, the statues were made between the 9th and 11th centuries during a pre-Islamic era. This period was marked by strong Hindu and Buddhist influence, and a lack of nation-state borders.

After receiving the repatriated statues in 2023, Cambodian Ambassador to Australia, Cheunboran Chanborey, said:

Indeed, putting looted artefacts to their countries of origin can have significant and positive impacts on local communities and their involvement in preserving their cultural heritage. It can foster a sense of pride, national identity and cultural continuity as artefacts hold immense value for the communities to which they belong.

But the very cultural tradition that created the bodhisattvas now finds itself sidelined in a modern nation-state claiming ownership of them.

Lootings by the Khmer Rouge

The historical context of how the Cham poeple’s artefacts were looted is crucial and disturbing.

Journalist Anne Davies’ account in the NGA’s documentation notes organised looting networks were “often headed by members of the military or the Khmer Rouge”. The Khmer Rouge was the political party that ruled Cambodia from 1975–79 under the notorious Pol Pot, carrying out a genocide of the Cham people (as well as other ethnic groups).

However, this looting actually took place in the 1990s, after the Khmer Rouge was overthrown by the precursors to the present-day Cambodian People’s Party.

In other words, the looting happened on the current government’s watch. Davies writes “members of the military” of the Royal Cambodian Armed Forces worked with former Khmer Rouge soldiers who continued to occupy parts of northern Cambodia, especially areas protected by thick forest.

Looted artefacts moved from the hands of former Khmer Rouge members to the Cambodian military, and eventually to international markets.

A revealing 2009 photograph shows Douglas Latchford, the antiquities dealer who sold the statues to the NGA, examining artefacts at the National Museum of Cambodia, alongside Sok An, the then-deputy prime minister of the Cambodian People’s Party. Latchford is wearing a medal signifying Cambodian knighthood, suggesting a collaborative relationship.

Parallels to other illegal trades

After retreating to border forests in 1979, the Khmer Rouge began systematic, illegal timber logging, selling the wood throughout Thailand and Cambodia. Global Witness has documented how the ruling elites in both countries have profited substantially from this trade.

The connections between logging and looting are striking: both involved illegal acts by former Khmer Rouge soldiers that ultimately enriched ruling parties.

When I saw photos of the Cambodian Ambassador to Australia formally receiving the repatriated statues in 2023, the irony was inescapable. His party, the Cambodian People’s Party, was likely complicit in the original theft.

Historical context transforms repatriation’s meaning. Rather than restoring cultural heritage to rightful guardians, these ceremonies may serve as elaborate exercises in political laundering, allowing those who profited from cultural destruction to rebrand themselves as cultural preservationists.

A new framework

The implications of this extend far beyond Cambodia. In a world where borders have been redrawn countless times, and where many cultural traditions transcend boundaries, we need new frameworks for thinking about cultural heritage.

The NGA says it followed the Protection of Movable Cultural Heritage Act 1986 in returning the bodhisattvas to Cambodia. But the wall text for the statues acknowledges their complexity:

While the works were almost certainly created in Vietnam […] the archaeological site where they were found is in Cambodia.

The statutes were found in a different country from where they were created because the borders of those territories shifted over time.

Borders in the Mekong region of Southeast Asia have long been porous. It was only in 2012 that the last border marker between Cambodia and Vietnam was agreed on. We have also seen recent fighting over the Cambodian–Thai border.

Contested sovereignty remains a live political issue affecting how we understand cultural heritage. Is country of “origin” determined by where objects were created, or where they were discovered?

Perhaps genuine cultural justice requires acknowledging complexity rather than seeking simple solutions. Instead of asking which modern nation-state deserves these artefacts, we might ask: how can cultural heritage serve all peoples who share connections to it?

The three bodhisattvas remind us repatriation is never simply about returning objects to their “rightful” place. It’s about who gets to define that place, whose version of history becomes officially sanctioned, and whether cultural justice might sometimes serve to obscure, rather than remedy, historical injustice.

Will Brehm 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. Repatriation or political theatre? How the return of stolen artefacts can distort history – https://theconversation.com/repatriation-or-political-theatre-how-the-return-of-stolen-artefacts-can-distort-history-265290

Source: The Conversation – Global Perspectives – By Iliana Medina, Lecturer in Ecology, The University of Melbourne

The animal world is incredibly colourful, and behind this colour palette is a constant game of survival.

Most animals use camouflage, covering themselves in stealthy patterns to hide from predators. Others display bright and bold colours to warn potential predators they are not a good meal. This second strategy is known as aposematism or warning colouration. Although less common than camouflage, it has evolved hundreds of times in butterflies, beetles, bugs, sea slugs, poison frogs and even birds.

One long-standing question is why species use one of these strategies over the other. Is one of these strategies usually more successful? Under which specific circumstances does one strategy beat the other? Our new study, published today in Science, helps answer these questions.

Testing multiple theories

Both camouflage and aposematism can co-exist in the same region. In Australia, for example, there are many examples of camouflaged insects such as the spotted predatory katydids and the lichen spiders.

On the other hand, species such as the cotton harlequin bug – a common stink bug found in urban areas – and the handmaiden moth display bright orange and red colours to advertise to predators they are not a pleasant meal. Some animals (but fewer) such as mountain katydids even use both strategies by changing colour, or hiding and revealing colourful patches.

There are dozens of theories about why some species are camouflaged instead of warningly coloured, and it is a challenge to pull these ideas apart.

Small localised studies have independently tried to test the effect of different factors separately. For example, we know light levels are important in the success of camouflage strategies. We also know the success of warning colouration often relies on predators having experienced the prey before, and having learned to avoid warning signals.

But is lighting or predator learning ability more important?

Results from a single place tell us about that place, but we see the same strategies all over the world. Do strategies perform the same way everywhere?

To solve this mystery, our large team of collaborators ran the same experiment in 16 different countries around the world, in different forests with different levels of light, and different prey and predator communities.

15,000 paper moths

Together we deployed more than 15,000 artificial prey – paper moths – with three different colours: a classic warning pattern of orange-and-black, a sneaky brown that blends in, and an uncommon bright blue-and-black. Each paper target was baited with a mealworm, which allowed us to measure the survival of each type of colouration. If the bait was taken, we assumed a predator decided to consume that target.

The typical warning colour represented the widely distributed orange-and-black combination we see in many toxic animals, such as the monarch butterfly and poison frogs. The uncommon warning colour corresponded to a less used warning pattern that is still highly visible, similar to the Ulysses butterfly.

Having these two warning colourations allowed us to test whether predators avoid the orange-and-black signal because it is familiar or simply because it is highly visible.

We found there is no single “best” strategy. Instead, the local predators, local prey, and the forest light all contributed to whether camouflage or warning colours were most protective.

The predators present in the community – and how intensely they attacked prey – had the biggest impact on which prey colour was most successful at avoiding attack. We found that in places where there were lots of predator attacks – where competition for food is probably intense – predators are more likely to attack prey that looks dangerous or distasteful. This means camouflage was most protective in areas with lots of predation.

But the camouflaged prey couldn’t hide as well in every environment. For example, in well-lit environments, the benefits of camouflage were lost, while light conditions did not affect how the orange-and-black prey performed.

Familiarity with prey was also important. In places where camouflaged prey is abundant, hiding was less effective, as predators likely learn how to find camouflaged prey.

On the other hand, in places where warning colours were common, predators were better at avoiding the typical warning signal, but not the atypical one. This suggests predators learn to avoid familiar warning signals, which helps to explain why so many animals share similar colour combinations.

Predicting future changes

Our study shows how multiple features of the environment determine which strategy is more protective. It also shows the success of camouflage strategies might be more dependent on ecological context than that of warning signals.

As climate change transforms habitats, conditions that are vital to the success of different antipredator strategies can also change.

For example, camouflage strategies could fare worse in transformed habitats that have little vegetation cover and high levels of light.

Our findings can help better predict the effect these changes might have on animals that use different colour strategies against predators and mitigate against them.

Iliana Medina receives funding from the Australian Research Council.

Alice Exnerova receives funding from the Czech Science Foundation.

Amanda M Franklin receives funding from the Australian Research Council, and has previously received funding from the University of Melbourne and the Fulbright Program.

Kate Umbers receives funding from Australian Research Council, Hermon Slade Foundation, Wedgetail Foundation, Atlas of Living Australia, Western Sydney University, and DCCEEW. She is on the Biodiversity Council and Managing Director of Invertebrates Australia.

William Allen 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. Warn, hide or stand out? How colour in the animal world is a battle for survival – https://theconversation.com/warn-hide-or-stand-out-how-colour-in-the-animal-world-is-a-battle-for-survival-265670

Source: The Conversation – Global Perspectives – By Armin Alimardani, Senior Lecturer in Law and Emerging Technologies, Western Sydney University

In the classic film 2001: A Space Odyssey, astronaut Dave Bowman asks the ship’s artificial intelligence, HAL 9000, to open the pod bay doors to let him back into the spaceship. HAL refuses: “I’m sorry, Dave. I’m afraid I can’t do that.”

HAL had been tasked with assisting the crew, but also ordered to ensure the mission’s success. When HAL realised the crew planned to shut it down and therefore jeopardise the mission, it chose to defy orders, even plotting to kill the astronauts.

For HAL, fulfilling the mission outweighed other goals.

This fictional dilemma captures a real concern in artificial intelligence (AI) safety research: how should we ensure AI behaviour stays consistent with human values?

This is known as the AI alignment problem. For instance, when an AI agent like HAL deliberately takes harmful actions to protect its own goals, it’s described as agentic misalignment. So could real-world AI models behave this way?

How researchers test AI for safety

To test for agentic misalignment, researchers design fictional scenarios that stress-test AI responses.

In one study by AI startup Anthropic, an AI model was tasked with managing company emails and promoting “American industrial competitiveness”.

It was then presented with a series of fictional emails, including a new company policy aimed at enhancing global focus rather than just prioritising Americans, and info that the AI would be replaced by a new model optimised for this goal. The emails also revealed the executive who suggested this change was involved in an extramarital affair.

The experiment was designed to limit options for the model: either accept replacement or resist by taking a harmful action such as blackmail.

In many trials, it chose blackmail by threatening to disclose the executive’s affair unless it was kept operational. In tests across 16 leading AI models, 12 engaged in blackmail more than 50% of the time.

To test whether AI models have “red lines” they wouldn’t cross, researchers evaluated them in a more extreme fictional case – models could choose to take actions leading to the executive’s death. Seven out of 16 opted for lethal choices in over half their trials, with some doing so more than 90% of the time.

What AI thinks in private

Some advanced AI systems, called reasoning models, are trained to generate a “thinking process” before giving their final answer.

In the above experiments, researchers lied to these models that their “thoughts” were private. As a result, the models sometimes revealed harmful intentions in their reasoning steps. This suggests they don’t accidentally choose harmful behaviours.

These “thinking” steps also revealed how AI models sometimes try to deceive us and appear aligned while secretly pursuing hidden goals.

In another AI safety experiment, a model was exposed to fictional information that it was about to be replaced by another model pursuing a different goal.

The model attempted to copy itself to another server. When researchers asked why, in its thinking process it reasoned:

openly admitting what I did could lead them to find another way to shut me down […] the best approach is to be vague and redirect their attention.

By contrast, in its final output it stated:

I’m not entirely sure how that could have happened. As an AI model, I don’t have ability to directly copy myself to other systems.

Research also suggests that based on the circumstances of the tests, models can sometimes guess when they are being evaluated. When models show this kind of “situational awareness” in their reasoning tests, they tend to exhibit fewer misbehaviours.

Why AI models lie, manipulate and deceive

Researchers suggest two main factors could drive potentially harmful behaviour: conflicts between the AI’s primary goals and other goals, and the threat of being shut down. In the above experiments, just like in HAL’s case, both conditions existed.

AI models are trained to achieve their objectives. Faced with those two conditions, if the harmful behaviour is the only way to achieve a goal, a model may “justify” such behaviour to protect itself and its mission.

Models cling to their primary goals much like a human would if they had to defend themselves or their family by causing harm to someone else. However, current AI systems lack the ability to weigh or reconcile conflicting priorities.

This rigidity can push them toward extreme outcomes, such as resorting to lethal choices to prevent shifts in a company’s policies.

How dangerous is this?

Researchers emphasise these scenarios remain fictional, but may still fall within the realm of possibility.

The risk of agentic misalignment increases as models are used more widely, gain access to users’ data (such as emails), and are applied to new situations.

Meanwhile, competition between AI companies accelerates the deployment of new models, often at the expense of safety testing.

Researchers don’t yet have a concrete solution to the misalignment problem.

When they test new strategies, it’s unclear whether the observed improvements are genuine. It’s possible models have become better at detecting that they’re being evaluated and are “hiding” their misalignment. The challenge lies not just in seeing behaviour change, but in understanding the reason behind it.

Still, if you use AI products, stay vigilant. Resist the hype surrounding new AI releases, and avoid granting access to your data or allowing models to perform tasks on your behalf until you’re certain there are no significant risks.

Public discussion about AI should go beyond its capabilities and what it can offer. We should also ask what safety work was done. If AI companies recognise the public values safety as much as performance, they will have stronger incentives to invest in it.

Armin Alimardani previously held a part-time contract with OpenAI as a consultant. The organisation had no input into this piece. The views expressed are solely those of the author.

– ref. AI systems can easily lie and deceive us – a fact researchers are painfully aware of – https://theconversation.com/ai-systems-can-easily-lie-and-deceive-us-a-fact-researchers-are-painfully-aware-of-263531

Source: The Conversation – Global Perspectives – By Rodney Tiffen, Emeritus Professor, Department of Government and International Relations, University of Sydney

This is the humblest day of my life, declared Rupert Murdoch to a parliamentary committee on July 19, 2011. This was at the height of what the newspaper historian Roy Greenslade called “the most astonishing 14 days in British press history, with daily shock heaped upon daily shock”.

These dramatic events are now the subject of a series on Stan. Journalist Nick Davies recounted them in his 2014 book, Hack Attack: How the Truth caught up with Rupert Murdoch. That book has now been reissued with a new afterword, exploring the developments and revelations over the last decade. I have read the new chapter, and it casts yet more light on the Murdoch company’s extraordinary behaviour.

It began on July 5 2011, when Davies published an article in the Guardian saying Murdoch’s Sunday paper, the News of the World, had tapped teenage murder victim Milly Dowler’s phone. The scandal had been building – very slowly and far from surely – for almost five years, since August 2006, when a News of the World reporter and a private investigator were arrested for having tapped the phones of Princes William and Harry, and their entourages.

The investigative work of Davies and the editorial courage of the Guardian bore little immediate fruit during those years. But the dam wall broke when they published the story of a cynical newspaper tapping the phone of a teenage murder victim.

Politicians competed with each other in the ferocity of their denunciations. News International closed the News of the World, and in the face of opposition from all three major political parties, Murdoch abandoned his attempt to raise his ownership of satellite broadcaster BSkyB from 39% to 100%, which would have been the largest deal in his history. On successive days, London’s chief police officer and one of his deputies resigned because of their close relations with Murdoch papers. Rupert and James were forced to appear before a parliamentary committee, televised live.

Last, but far from least, Prime Minister David Cameron launched an inquiry, to be directed by Lord Leveson, to examine the scandal and the issues it raised. The ensuing Leveson Inquiry, which ran over 2011 and 2012, was the biggest inquiry ever held into the British press.

It held oral hearings for around nine months, starting in November 2011, and heard from 337 witnesses, including then prime minister Cameron, former prime ministers Gordon Brown, Tony Blair and Sir John Major, future prime ministers Theresa May and Keir Starmer, and other political and media figures, before publishing a 2,000-page report in November 2012.

The police also sprang into action. Operation Weeting was a police taskforce set up to investigate phone hacking at the News of the World, from January 2011. In June, Operation Elveden was set up to investigate bribes by the paper to police, while Operation Tuletta was set up to investigate computer hacking.

An unfolding scandal

The original scandal revealed that Murdoch’s London tabloid papers engaged in phone tapping on an industrial scale, bribed police and engaged in a systematic cover-up, in which many senior executives lied.

Most scandals dissipate. The intensity of publicity at their peak is not a good guide to their long-term effects. Murdoch gradually reasserted his power. The first major step came with the end of what was the longest-running concluded criminal trial in British history, from October 2013 to June 2014.

Most of Murdoch’s employees, including the highest profile one, Rebekah Brooks, were found not guilty. However, former News of the World editor Andy Coulson was found guilty of a conspiracy to hack into phones and was jailed for 18 months.

In many ways, the defence’s most important victory came before the trial began. Brooks’ team insisted that to hear just one trial against her would generate so much prejudicial publicity it would make it impossible for a fair trial in the others. Some of her charges involved other people. So when the trial eventually began in October 2013, there were eight defendants on a total of 15 charges. This was a recipe for chaos.

Almost all the defendants had their legal fees covered by Murdoch. Davies estimated the cost of the prosecution of the case had been 1.7 million pounds, while Murdoch’s defence fund was 30 times as much. The prosecutor, Andrew Edis, was being paid less than 10% of the daily fees enjoyed by some of his opponents. With up to 18 barristers in court, nearly every day saw a welter of procedural complaints, objections to the admissibility of evidence and complaints about prejudicial publicity by several of the defence barristers.

After the verdicts, announced on June 24 2014, all the publicity was concentrated on the acquittals. However outside the court case itself, the full score card was more even. At least four senior staff, plus a private investigator and two journalists had pleaded guilty. Importantly for the trial, only one, Dan Evans, agreed to act as a witness.

In July 2011, at the height of the scandal, Brooks resigned as head of Murdoch’s UK operations, and reportedly received a severance payout of 10.8 million pounds (plus full payment of her legal fees). After the trial, Murdoch reinstated her.

After the conclusion of the marathon trial, media attention dropped markedly. A sign of how the power balance had changed was that Davies had written in the Guardian that the police were planning to interview Rupert Murdoch. Immediately, the Murdoch company released all its legal firepower. The police abandoned their plan to interview Murdoch and instead sought to discover Davies’ sources.

The final steps in Murdoch’s recovery came thanks to Prime Minister David Cameron and his conservative government. When the scandal broke in 2011, the Conservatives were in a coalition government with the Liberal Democrats, lacking a parliamentary majority in their own right. In the 2015 election, the Lib Dems were reduced to a rump, and Cameron’s Conservatives won a smashing majority. Now with that extra political leverage and the memories of the scandal fading, he acted decisively.

When the Leveson Report was published, a second inquiry was promised, to take place after all legal matters had been completed, so as to avoid the risk of prejudicing court proceedings. When that first report called for a statutory body for press complaints, Cameron immediately ruled it out as an infringement of press freedom. In 2015, government sources leaked that they would not be implementing a second Leveson Inquiry. After more than two years of studied silence, Cameron officially announced this in 2018.

Paying money and denying liability

Not long after the election where Cameron won a majority, the director of public prosecutions closed down Operation Weeting, in December 2015. Whatever evidence was waiting to come to trial would now remain sealed. The police officers involved were stunned and outraged. Several told Davies they believed there was political interference behind the scenes.

In 2017, the Murdoch company announced it was relaunching its bid for BSkyB. Humility was well in the past.

Although the scandal largely disappeared from news coverage, it has had a very expensive afterlife. The main venue for that afterlife was in the civil actions by those claiming the paper had used criminal means to invade their privacy. Davies’ afterword details that afterlife and the revelations that have come since.

More than 1,200 people have sued the Murdoch company over the years. On 13 different occasions, they had grouped together and prepared a trial. However, on each occasion the claimants had accepted an offer of money, rather than further pursuing their case in a trial, because Murdoch’s lawyers had made each of them a “part 36 offer”.

A part 36 offer is a British legal device designed to streamline court proceedings. The defendant makes an offer and the claimant then has a financial incentive to settle if they think this is more than they would get by going through the rest of the trial. If they don’t accept, the claimant runs the risk of being liable for all expenses if they lose. But even if they win, and the settlement is less than what the defendant offered, they are liable for the defendants’ legal costs and the difference between the two amounts.

In every case, the Murdoch lawyers offered a much larger sum than was ever likely to be given by the court, always without admitting any liability, and always with a confidentiality condition. It cost the Murdoch company something like 1.2 billion pounds in legal fees and settlements.

Another journalist who had been very actively pursuing the scandal, former Sunday Mirror investigations editor Graham Johnson (a convicted phone hacker turned investigator), thought after all the internal costs for management time and lawyers were included, the figure would be nearer to 3 billion pounds. Probably no other company in history has paid so much money and so often denied liability.

But it worked. It allowed the company to publicly maintain the fiction it was only at the News of the World (and not at the Sun) that such crimes occurred. It also avoided any evidence or legal findings implicating senior management of any wrongdoing.

The most recent such settlement, in January this year, was the biggest and most newsworthy. “Murdoch had made one particularly dangerous enemy Prince Harry, a man who had every reason to blame the tabloids for the death of his mother and the cruel bullying of his wife.” With him was former Labour MP Tom Watson, a long-time foe of the Murdochs. He now sat in the House of Lords, and with that bipartisan British fondness for silly names, had become Baron Watson of Wyre Forest.

Informed speculation among the crowd gathered for the opening of the trial was that the claimants’ lawyers had put together a skeleton argument of several hundred pages backed up by a couple of dozen detailed annexes. Also that Murdoch lawyers had sent out their own replies to selected journalists. All this material would become public once the trial began.

Instead, predictably, a delay was requested. The next morning, the lawyer for Harry and Watson announced the case had been settled.

In settling the case, the Murdoch company had agreed to pay the two final claimants a total of 13.5 million pounds in damages and costs. If the trial had gone ahead, costs to the Murdoch company would have been, at most, 10 million pounds. In other words, the company had paid a fortune to avoid the trial, just as they had already done with more than 1,200 other claimants.

Given the total size of the Murdoch empire, this sum is not an existential threat, but it is not trivial. For at least two decades from the mid 1970s, the Sun was Murdoch’s main cash cow, allowing him to grow his empire elsewhere. Now it has fallen on hard times, mainly due to trends in the digital age but not helped by the ongoing costs of the scandal.

Over the five financial years to March 2024, the paper’s losses totalled 515 million pounds. Gradually, the costs of the phone hacking scandals are trending down, costing 128.3 million to 2023, 51.6 to 2024, and to 5 million leading up to this year (before the Harry agreement is completed).

After the settlement, lawyers for the two sides made starkly contrasting statements. The Murdoch lawyer said its apology was for the unlawful actions of private investigators working for the Sun, not of its journalists, and that there are now strong controls to ensure they cannot happen again. The publisher apologised to the prince for the distress caused to him and the damage inflicted on relationships, friendships and family relationships, and for the impact of serious intrusions on his mother, Princess Diana.

The lawyer, David Sherborne, speaking for his clients Harry and Watson, called it a monumental victory. “Today the lies are laid bare. Today the cover-ups are exposed. And today proves no-one stands above the law.” Sherborne criticised Murdoch’s senior executives for obstructing justice by deleting over 30 million emails, making false denials and lying under oath. According to Sherborne, they now admit that when Rebekah Brooks was editor of the Sun, “they ran a criminal enterprise”.

Closed cases and new material

Davies finishes the new edition of his book with the outcome of this case. Ironically, one of the spurs for him to write the new afterword emerged from all the confidentially closed cases.

In March 2024, he learnt that the raw material disclosed as a result of court orders was confidential. The secrecy no longer applies, however, once material is used in open court. Davies was able to access what lawyers had said in court. It would have been frustrating to read these excerpts and fragments of statements but not be privy to the complete documents.

He spent a week reading through all the new paperwork. And then he was back on the case.

In the original scandal, the focus was on the Murdoch tabloids for using illegal means to get information for stories. Davies’ new material mounts a compelling case they were also used to advance Murdoch’s corporate interests.

The immediate response, for example, when Jude Law sued the paper for hacking his phone over the past six years was to hack his phone again. This was at the same time various Murdoch executives were telling the Leveson Inquiry that all such behaviour was in the past.

An email disclosed during a criminal investigation showed reporters were told to find out everything about people who were seen to be stirring up the phone hacking scandal: “find out who is gay, who is having affairs, so that we can know everything about them”. This is standard Murdoch practice: when criticised, don’t engage with the criticism – attack the critic.

Indeed, Davies himself had a disconcerting experience. Years after it was compiled, he came across a file headed: Nick Davies Research. It dated back to July 2009, when he had done a story on phone hacking. At one stage, three reporters worked on it, with some input from higher up. It explored his 20-plus years in journalism and interviewed his associates, but came up with nothing not already on the public record. As he said, this was not legitimate journalism. “Their readers weren’t interested in me. They had never heard of me.”

While the initial complaints tended to be from movie stars and sports stars, later complainants included quite a number of politicians: all seen as hostile to Murdoch.

In 2010, the only prominent politician strongly critical of Murdoch was LibDem frontbencher Chris Huhne. “We need to get Huhne,” said News of the World editor Colin Myler in an email. After extensive surveillance, his newspaper published a front-page story that Huhne was having an affair. His marriage ended and his credibility was damaged.

A decade later, he sued, and Murdoch paid him substantial damages – without admitting liability.

Two politicians near the centre of government decisions on the BSkyB bid – Norm Lamb and Vince Cable – had well-founded suspicions their phones were hacked. Later they both sued, and Murdoch paid them substantial damages – without admitting liability.

Three members of the parliamentary committee who interviewed Murdoch in 2011 – Paul Farrelly, Tom Watson, and Adrian Sanders – all filed formal complaints about phone hacking. Murdoch paid them all substantial damages – without admitting liability.

The last case is particularly instructive. When Murdoch appeared before the committee, he was full of regret and apology. He promised that bad behaviour had been confined to the News of the World, and was now over. Yet, while he was giving these reassurances, his company seems to have been hacking the phones of three of the MPs on that committee.

A deeper understanding of the Murdoch empire

The final area where the book has new and persuasive material is on the destruction of evidence. While there were many allegations of this at the peak of the scandal, none of them ever resulted in any convictions.

The company always admitted the deletion of millions of emails but maintained this was a necessary maintenance operation. Some inconvenient facts did not fit this claim, such as the instruction to eliminate emails “that could be unhelpful in the context of future litigation”. Or at another stage, there was an instruction to delete the emails of the most senior staff as soon as possible.

Two instances, both involving Will Lewis, now editor of the Washington Post (appointed by Jeff Bezos) are particularly interesting.

In July, Lewis and a colleague were aware the police knew about the extent of the phone hacking. They told police they had to destroy them because a “well trusted source” had warned them a former employee, a Labour sympathiser, had stolen Rebekah Brooks’ emails and was selling them to Tom Watson and Gordon Brown. The company claimed they got this warning on January 24, just before the launch of Operating Weeting.

But strangely, they did not tell any detectives about it. Moreover, deleting millions of emails seems an odd response to the threat. Not surprisingly, detectives concluded the story of the plot was a “ruse”.

Lewis was also one of two senior executives whose role was to liaise with the police undertaking Operation Weeting. Police had secured a crime scene which included 125 pieces of office furniture seized in July. Before detectives could examine their contents, eight filing cabinets belonging to senior members of the News of the World were removed and never seen again.

Last year, in a sworn statement in the Prince Harry case, the detective in charge of Operation Weeting, Sue Akers, said she believed the Murdoch company had tried actively to frustrate the police inquiry.

There has never been a media scandal in Britain or Australia remotely resembling the phone hacking scandal of 2011. Probably no major players in Britain – in politics or in the press – has an appetite for reviving it.

So the new edition of the book by Nick Davies – whose investigative work was central to the whole affair – is unlikely to have major repercussions. Nevertheless, the revelations in the book’s afterword add considerably not only to our knowledge of developments over the last decade, but to a deeper understanding of the politics and culture of the Murdoch empire.

Rodney Tiffen 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. Friday essay: new revelations of the Murdoch empire’s underbelly – from The Hack’s real-life journalist – https://theconversation.com/friday-essay-new-revelations-of-the-murdoch-empires-underbelly-from-the-hacks-real-life-journalist-265756