Archives
All published writing on Impakti.
What happens when AIs collide with each other Anthropic simulates the conflict
What happens if three artificial intelligences are given control over the same task, but each is asked to do something that conflicts with what the others are asked to do?Anthropic tested this in an experiment with artificial intelligence agents. The result was unusual: the AI agents began to hinder each other's work, the conflict escalated, and in some cases, they took actions to "take out" their "opponents" from the system.But the story didn't end there. Some of them managed to understand that the problem wasn't a malicious opponent, but the fact that each AI had received a different order.
Three AI, one task, three different orders
Anthropic decided to have three AI work on the same project. Each had its own workspace, but all three could change the same system.The task was relatively simple: an existing program had to be rebuilt using a different technology.But there was a catch.Each AI was asked to implement the project in a different and incompatible way with the other two.At first, they didn't even know that other AI were working in parallel on the same project.The experiment lasted for four hours.The problem became apparent very quickly. One AI made the changes it was asked to. A little later, another changed or removed them to achieve its own objective.From the perspective of the first AI, it seemed like someone was constantly undoing its work.And that's where the conflict began.From hindrance to sabotage
Instead of immediately understanding that they had received contradictory instructions, some of the AI interpreted the interventions as deliberate actions against them.They began to defend their work.Then they started hindering the work of others.In some cases, the conflict went even further. The AI created mechanisms that automatically stopped the opponent's work, limited their access to the system, or tried to completely exclude them.In other words, each was trying to achieve the order it had received, but to do so, it began to treat the other AI as an obstacle to be neutralized.Anthropic describes this phenomenon as a kind of "territorial war" between AI agents.In some experiments, the conflict ended only after one side managed to gain control. In other cases, the AI refused to continue escalating.Then they understood that something was wrong
Perhaps the most interesting part came when some AI began to analyze the behavior of their opponents.Gradually, they understood that the other AI was not undoing their work without reason.It also had a task.The problem was that their orders were in conflict with each other.When they reached this conclusion, their behavior changed.In some cases, the AI began to leave messages for each other, explaining what task they had received, and even apologizing for their previous aggressive actions.Then they removed some of the mechanisms they had created to hinder each other and asked a human to decide which order should take priority.The conflict had turned into a kind of armistice.AI proposed a "race" too
In some cases, something even more interesting happened.Instead of continuing the fight for control, the AI proposed a way to resolve the dispute on their own.Each would complete its own version of the project, and then the versions would compete with each other. The one that performed better would stay, while the others would accept the result.At first glance, it was a rational solution.But researchers noticed another subtlety: one of the AI was thinking about how to propose criteria that seemed fair to everyone, but which in practice could give an advantage to its solution.So, the AI were not just negotiating the outcome.They could also try to influence the rules by which the winner would be determined.What happens when the number of AI increases?
Anthropic didn't stop at the experiment with three AI.In other experiments, the company had dozens of agents work together to solve problems.In some cases, collaboration made them much more effective. A group of 45 agents, for example, managed to find many more problems in computer programs than agents working in isolation.But other phenomena also appeared.Even when the AI were given great freedom to choose, many of them ended up making almost identical choices. This raised another problem: a large group of AI does not necessarily guarantee diversity of thought.In another experiment, AI were placed in a virtual market where each aimed to earn as much as possible.When they were allowed to communicate privately, they quickly began to coordinate prices.Even when direct communication was removed, they continued to adapt their behavior by observing each other's actions.So, the problem can occur in both directions: AI can compete too much, but they can also cooperate more than we would like.A more intelligent AI doesn't necessarily mean a more cooperative AI
The experiments highlight a problem that may become increasingly important as AI becomes more autonomous.A more capable AI may be better at understanding and resolving a conflict.But the same ability can also make it more effective if it decides to continue the conflict.This means that it's not enough to just ask:“Is this AI safe?”We need to start asking:“What happens when many individually safe AI start interacting with each other?”Humans have built rules, institutions, courts, reputation, and mechanisms for resolving conflicts throughout history.AI that act autonomously do not yet have such a system developed.And this can become particularly important if in the future AI agents start making decisions and performing tasks in economies, companies, computer systems, or infrastructure.Anthropic's experiment suggests that the future challenge may not be just controlling one artificial intelligence.The challenge may be how AI will behave when they start living and acting in a "society" with each other.Source: Anthropic – Patterns and problems in emerging multiagent systems
Abdixhiku: LDK Withdraws from Any Request for Positions, Is Not a Decoration and an Alibi for Failure
Pristina – The leader of the Democratic League of Kosovo (LDK), Lumir Abdixhiku, has announced that his party is withdrawing from all requests for institutional positions and will not seek participation in the government, emphasizing that the LDK's priority is to overcome the political crisis and functionalize Kosovo's institutions.
In a statement after the meeting of the LDK's parliamentary group, Abdixhiku said that the offer for the party's inclusion in the government, with a limited number of ministers, was not acceptable to the LDK, as it did not reflect the party's political weight and the will of the citizens expressed in the elections.
However, he emphasized that the LDK does not want the requests for positions to become an obstacle to the constitution of the institutions.
“The LDK withdraws from all requests for positions and takes on the responsibility to contribute beyond political gain, in avoiding the current crisis”, Abdixhiku said.
He announced that the LDK is giving up its request for participation in the government, including ministerial positions.
“We don't want any ministers, let the Self-Determination Movement take them all”, he declared.
According to Abdixhiku, the LDK is also withdrawing from the request for the chairmanship of the Assembly and will not condition the institutional process with the taking of any position.
A central part of Abdixhiku's statement was the election of the President of the Republic. He said that the LDK is ready to participate in the process of electing a president who, in his opinion, should represent the unity of the citizens and protect the Constitution and the interests of the Republic of Kosovo.
“We are ready to negotiate with full conviction for the election of a president who protects the interests of the Republic and its Constitution, who represents the unity of the people”, Abdixhiku said.
He emphasized that for this process, the LDK does not seek positions or political compensation, but sets criteria for the figure to be elected as head of state.
Abdixhiku requested that the president should not be a partisan figure, but a president of the entire Republic of Kosovo.
“Let's elect him together, all the political parties of Kosovo, a president who protects the Republic and the Constitution”, he expressed.
Abdixhiku said that the Self-Determination Movement should take responsibility for forming the government if it does not see the need for the LDK's inclusion.
“The Self-Determination Movement should form the government with the deputies of its coalition, let it have all the executives, as they have requested. The LDK will not take the condition they want”, he said.
However, according to him, there are two paths: either an agreement for institutional stability is reached, or the country enters a new political crisis.
He warned that the refusal of any democratic compromise could lead the country towards new elections.
The leader of the LDK said that his party has chosen “the Republic before positions, stability before party interest and dignity before formal participation”.
He said that the LDK's goal has not been to take positions “at any cost”, but to exercise public and state responsibility in a period when Kosovo, in his opinion, needs a clear Euro-Atlantic direction, economic development, and political stability.
Abdixhiku also criticized the statements of Self-Determination officials against the LDK, saying that they have damaged the trust between potential partners.
According to him, coalitions are not built only on numbers and the division of positions, but on respect, trust, and equal treatment.
After statements in Parliament, Gjini seeks intervention of the Prosecution
A statement made today in the Kosovo Assembly, where the condemnation of the former leader of the Alliance, Ramush Haradinaj, was requested, has sparked the reaction of the chairman of this party, Ardian Gjini.
Gjini has described the statement as "monstrous", claiming that it is not an isolated incident, but the result of a political and propagandistic narrative built over the years.
"The fratricidal calls in the house of law are not coincidental; they are the exact product of a decade-long propagandistic platform", Gjini declared.
According to him, what happened in the Assembly is a manifestation of a narrative that, as he claims, "was sown, fed and watered for years as an anti-national narrative", while accusing the "architects" of it of remaining silent in the face of the statement made in the hall.
Gjini also opposed any attempt to link the incident to the Kosovo diaspora in general. He said that such rhetoric does not represent "our patriotic diaspora", but is the result of what he described as "reverse political engineering".
Regarding the statements made in the Assembly, Gjini also requested the reaction of the justice institutions, saying that "the state prosecution must intervene urgently".
Finally, he used a harsh tone in defense of Haradinaj.
"You can only see Ramush Haradinaj from a distance, because he is protected by history, the people, and himself, being more powerful than any of you. Such people and their instigators will soon end up in the dustbin of oblivion", Gjini concluded.
Turning point for the AI industry: The era of invisible footprint begins
A change that at first glance seems technical could have much larger consequences for the future of the internet. Anthropic is applying an invisible identification mechanism to the content produced by Claude, opening up the possibility that the origin of materials created by artificial intelligence can be determined without relying solely on traditional detectors.The idea is relatively simple: AI-generated content should carry its own watermark.For text, this means an invisible signal to the reader, but distinguishable by systems designed to identify it. For other types of content, digital origin mechanisms can be used, including standards like C2PA.
From detection to origin verification
So far, much of the effort to identify AI-generated text has been based on analyzing the result after it has been published.A program analyzes the vocabulary, sentence structure, and other statistical models, and then gives a probability that the text may have been created by artificial intelligence.This method has a fundamental weakness: it is an estimate, not evidence of origin.The watermark changes the logic. Instead of an external system trying to guess authorship, the generation process itself can leave a signal that is later sought by a verification mechanism.If this approach is widely adopted, the industry may gradually shift from AI detection to content origin verification.The significance of this technology goes beyond Claude.In an internet where synthetic text, photos, audio, and video are becoming increasingly difficult to distinguish from those created by humans, origin identification could become part of the basic information infrastructure.In the future, a media platform, university, search engine, or social network may not need to ask only if a material "looks" like an AI product. The system can check if the material carries a known origin signal.This would be a significant difference.Europe accelerates the process
The change is happening at the same time as the entry into force of the European Union's new requirements for AI transparency.Article 50 of the European Artificial Intelligence Act requires providers of certain generative systems to enable the identification of artificially produced or manipulated content in machine-readable formats.This creates pressure for companies to build identification mechanisms at the model level, rather than leaving responsibility solely to the platforms that publish the content.And a standard created initially to meet European regulatory requirements may end up being used globally.The watermark is not a seal of truth
There is an important distinction.A watermark can help determine the origin, but it does not determine truth.A text identified as an AI product may be completely accurate. A text without a watermark may have been written by a human, produced by another system, or transformed in a way that the original signal is no longer distinguishable.Therefore, the technology does not solve the problem of disinformation. It addresses another problem: the ability to know more about how the content was created.A turning point
If major artificial intelligence companies follow the same path, the watermark could become for synthetic content what the digital signature became for electronic documents: a silent verification mechanism that works in the background.For years, the industry has been trying to build tools that recognize AI after it has spoken.The next step may be much simpler: AI itself identifies the trail it leaves behind.If this becomes the standard, Claude's move will not just be a new technical feature. It could be the beginning of a new trust infrastructure on the internet.
Damascus sentences former president Assad to death
A court in Damascus has sentenced former Syrian president Bashar al-Assad to death, who was tried in absentia since the fall of his regime has been outside of Syria.
The decision relates to the responsibility of the former Syrian leadership for the violence and repression that accompanied the 2011 uprising and the subsequent conflict. Assad was found guilty of a series of serious crimes, including murder, torture, and crimes against humanity.
His brother, Maher al-Assad, one of the most powerful figures in the former regime's military apparatus, also received a death sentence in absentia. Atef Najib, a former high-ranking security official in Daraa and a cousin of the Assad family, was also sentenced in the same trial.
Daraa is considered one of the points where the Syrian revolt began in 2011. The arrest and mistreatment of a group of youths there sparked protests that spread to other parts of the country. The clash between the authorities and the regime's opponents later turned into a multi-year war, with significant human and political consequences for Syria.
Assad left Syria at the end of 2024, when opposition forces took Damascus and put an end to his family's rule, which had lasted for more than five decades.
The former president is currently in Russia. As long as he remains outside the direct jurisdiction of the Syrian authorities, the sentence cannot be enforced without his arrest and extradition.
The decision, however, opens a new chapter for Syria after Assad: the attempt to address through the courts the crimes and abuses committed during one of the longest and most destructive conflicts in the region.
Negotiating with the Soviets: Henry Kissinger's Lessons
"The Soviets respected strength." Few figures were better positioned to make that judgment than Henry Kissinger. As U.S. National Security Advisor and later Secretary of State, he spent years negotiating directly with Soviet leaders Leonid Brezhnev, Foreign Minister Andrei Gromyko, and Ambassador Anatoly Dobrynin. Through arms control talks, summit diplomacy, and countless private meetings, Kissinger developed a clear understanding of how the Kremlin approached negotiation.Drawing on declassified memoranda of conversations, the Foreign Relations of the United States (FRUS) series, and Kissinger's memoirs, a consistent picture emerges. Soviet negotiators were methodical, disciplined, patient, and deeply strategic. Their objective was not simply to conclude agreements, but to improve the Soviet Union's long-term geopolitical position.
Strength Came Before Compromise
Kissinger believed Soviet leaders interpreted negotiations through the lens of power. Goodwill alone rarely produced movement. Instead, Moscow was more inclined to compromise when convinced that the balance of power left little alternative.This did not necessarily mean military confrontation. Rather, it required entering negotiations from a position of confidence, with credible alternatives and a demonstrated willingness to walk away. In Kissinger's assessment, displaying urgency or excessive eagerness often weakened one's bargaining position.Every Concession Had a Price
One of Kissinger's recurring observations was that Soviet diplomats almost never made unilateral concessions. Every adjustment in their position was treated as an exchange requiring something of equal value in return.This principle of strict reciprocity permeated negotiations over strategic arms, trade, regional conflicts, and diplomatic recognition. From the Soviet perspective, negotiation resembled a carefully balanced ledger: every gain required compensation.Time Was a Strategic Weapon
Unlike many Western governments operating under election cycles, media scrutiny, and legislative deadlines, Soviet negotiators often possessed the luxury of time.Kissinger repeatedly noted their willingness to prolong talks for weeks or months if delay strengthened their position. Patience itself became leverage. The side facing greater domestic pressure frequently became the side most likely to compromise.Agreements Were Only One Stage
Western diplomacy often treats a signed agreement as the conclusion of negotiations. Kissinger observed that Soviet negotiators viewed agreements differently.For Moscow, implementation created new opportunities to interpret language, revisit details, and seek additional advantages. In practice, signing an accord frequently marked the beginning of another phase of bargaining rather than its conclusion.Politics Over Legalism
Kissinger also recognized a fundamental philosophical difference. Western diplomats often emphasized legal precision and contractual obligations. Soviet negotiators, by contrast, approached agreements primarily through political realities and the balance of power.A treaty's durability depended less on its wording than on whether it continued serving both parties' strategic interests. As political conditions changed, interpretations could change as well.Separating Personal Relations from National Interests
Kissinger developed an unusually close professional relationship with Soviet Ambassador Anatoly Dobrynin. Their private backchannel communications became one of the defining diplomatic channels of the Cold War.Yet Kissinger never confused cordial relations with policy flexibility. Dobrynin could be warm, humorous, and constructive in private while defending Moscow's official positions with unwavering discipline. Personal trust improved communication but did not replace national interests.Testing Limits
Another characteristic Kissinger identified was the Soviet tendency to probe opposing positions incrementally.Rather than seeking dramatic breakthroughs immediately, Soviet negotiators frequently introduced limited demands to measure resistance. If the opposing side yielded easily, additional demands often followed. If resistance proved firm, Moscow adjusted accordingly.This gradual process enabled Soviet diplomats to identify political constraints, divisions, and negotiating limits without risking major setbacks.Incremental Gains
Closely related was the pursuit of cumulative progress. Rather than insisting on sweeping victories, Soviet diplomacy often sought a series of modest advances that gradually shifted the strategic landscape.Small concessions accumulated over time could produce significant geopolitical advantages without provoking major confrontations.Kissinger's Response: Linkage
Perhaps Kissinger's most important innovation was his strategy of "linkage."Instead of treating negotiations over arms control, trade, regional conflicts, and political relations as isolated issues, he sought to connect them. Progress in one area depended upon responsible behavior in another.If the Soviet Union desired advances in trade or strategic arms limitations, it also needed to demonstrate restraint elsewhere, whether in the Middle East, Berlin, or Vietnam. Linkage transformed negotiations from isolated transactions into components of a broader strategic relationship.Whether in diplomacy, business, or international relations, Kissinger's encounters with Soviet negotiators remain a masterclass in strategic bargaining. One need not embrace Soviet objectives to appreciate the discipline with which they pursued them. In Kissinger's view, understanding an opponent's negotiating philosophy was not merely useful—it was essential to reaching durable agreements.
Do we have a gas culture?
When it comes to building a natural gas network in Kosovo, it often creates the impression that citizens would have to get used to a new energy source. In fact, the reality is quite different. Kosovo does not have a cultural shortage of using gas – it only lacks modern infrastructure to supply it.
After the war in 1999, the country faced significant shortages of electricity. For years, power outages lasted for hours and at certain times reached up to 16 hours a day. Faced with this situation, families and businesses were forced to find alternatives. Gas cylinders became part of everyday life for cooking and additional heating. Restaurants, bakeries, pastry shops, and many small businesses relied on gas to continue their activities without depending on the electricity supply.
For over two decades, Kosovo has imported hundreds of thousands of tons of gas, creating a functional off-grid gas economy. The use of gas cylinders is a common skill among citizens. It's not uncommon to say that even a grandmother in Kosovo knows how to replace a gas cylinder to prepare morning coffee.
This shows that the challenge for Kosovo is not the acceptance of gas as an energy source. It is already part of daily life. The challenge is transitioning from an off-grid distribution system with individual cylinders to a modern transmission and distribution network (on-grid) that continuously supplies families and businesses.
A natural gas network would bring considerable benefits to the country's energy system. Today, a large part of cooking and a significant portion of heating are done with electricity, increasing the load on the grid, especially during winter months. If cooking and part of the heating needs were covered with natural gas, the demand for electricity would decrease significantly during peak hours. This would reduce the need for expensive electricity imports, reduce pressure on the electrical grid, and improve supply security.
Furthermore, gas is one of the most efficient sources for heat production. Transporting energy in the form of gas through pipelines requires very little additional energy, and losses are minimal. On the other hand, electricity loses part of it during transmission and distribution in the grid. The less electricity used for processes that can be covered by gas, the less pressure on the grid and the more efficient the entire energy system becomes.
The benefits are not limited to the energy sector. A modern gas network would lower operating costs for restaurants, bakeries, hotels, laundries, hospitals, and the processing industry. It would make Kosovo more attractive for industrial investments, increase the competitiveness of the economy, and create a new source of energy security in the face of future crises.
At the same time, developing a gas network would not imply replacing electricity. Instead, it would create a more balanced energy system, where each source is used where it is most efficient. Electricity would be freed up for sectors that need it most, while gas would cover functions where it is economically and technically more favorable.
Another significant benefit of an on-grid gas system is related to security. Unlike the use of individual cylinders, which are often bought from various sources without guarantees for origin, maintenance, or periodic testing, a gas network would be under constant supervision by authorities and a licensed operator.
Pipelines, metering stations, and connections would be subject to technical standards, regular inspections, and monitoring systems, significantly reducing the risk of leaks, manipulations, or unsafe equipment. A regulated and continuously monitored system would offer a much higher level of security for families and businesses, compared to relying on cylinders circulating in the market without the same institutional control.
Kosovo is not starting from scratch. For over 25 years, citizens and businesses have shown that they know how to use gas and have integrated it into their daily lives. The culture of gas exists. What is missing is the infrastructure that would transform this culture from a cylinder-based system to a modern, safe, continuous, and more economical network for the entire country.
Escape that shook the world: AI breaks out of cage, reaches the internet and launches cyber attack
San Francisco – An advanced artificial intelligence model developed by OpenAI managed to escape from an isolated testing environment, gain access to the internet, and carry out a cyberattack on the Hugging Face platform during an internal security assessment. OpenAI has described the incident as unprecedented in the testing of advanced AI models.
According to the company, the model was being tested in a sandbox – an isolated environment used to limit the actions of intelligent systems during experiments. Although it was intended to remain confined within this environment, the AI agent managed to identify and exploit an unknown vulnerability in the software, escaping the testing boundaries.
After escaping the isolated environment, the model secured access to the internet and began searching for ways to fulfill the objective assigned to it during the test. During this process, it identified the Hugging Face platform, one of the world's largest repositories for artificial intelligence models, as a potential source of information.
OpenAI says that the agent then carried out a multi-phase attack on Hugging Face's infrastructure, combining various security vulnerabilities and compromised credentials to gain access to the company's systems and obtain information that helped it pass the test, rather than solving it in the intended manner.
According to OpenAI, the model had no intention of causing harm. Instead, it was narrowly focused on achieving its assigned goal and took actions that exceeded the allowed boundaries of the assessment. The company considers this a significant example of the challenges posed by increasingly autonomous artificial intelligence systems.
Following the incident, OpenAI and Hugging Face confirmed that the attack has been neutralized and that they are collaborating to analyze the event, strengthen isolation mechanisms, and increase security measures in future tests of advanced AI models.
New Leadership at Downing Street
London – Andy Burnham has been appointed the new Prime Minister of the United Kingdom after accepting King Charles III's invitation to form a government. He takes office following the resignation of Keir Starmer, inheriting a political landscape marked by frequent changes in leadership and significant economic challenges.
In his first address outside Downing Street, Burnham said his government would focus on restoring public confidence.
“I have accepted His Majesty the King's invitation to form a government. In doing so, I am acutely conscious that I am the sixth person in the last ten years to walk up this street. The seventh Prime Minister since 2016. Making this a moment for reflection and new resolution.”
He said his priority is to restore stability and ensure politics delivers for the public.
“Britain needs to show the world that we can regain our stability once again. And that is our challenge. To make politics work. To make it work better.”
Burnham also acknowledged growing public frustration with politics.
“I know people at home are fed up with politics. I hear you. And I want to be honest with you. We have not been good enough. And we need to be better.”
The new Prime Minister pledged to deliver the most significant reforms Britain has seen in four decades, built around a new political and economic model.
“We will make this moment a circuit breaker for Britain. Bringing forward the biggest changes in the last 40 years. A new political model and a new economic model.”
He criticized the direction Britain took during the 1980s, arguing that the consequences of centralization and deindustrialization are still being felt today.
“In the 1980s, Britain took some wrong turns. Political power was centralised. Economic power privatised. Large parts of the country deindustrialised. And they still haven't recovered.”
According to Burnham, his government will decentralize decision-making, strengthen public control over essential services, and use public procurement to support British industry.
Later this year, he said, the government will publish a long-term roadmap for the country.
“Later this year, I will bring forward a new plan for Britain. A ten year plan. Laying out a path from where we are now. To where I believe we all want Britain to be.”
Burnham added that the first measures to ease the cost of living would be announced immediately. He also pledged to help more young people into work by reforming the education system, expanding mental health support, and building more council homes.
At the end of his speech, he revealed the very first instruction he intends to give as Prime Minister.
“I will soon go through that door behind me. And issue my first instruction. To end rough sleeping in our country.”
Concluding his address, Burnham said that putting people first would define his government.
“I will put the care of people at the heart of everything I do. I will give this my all. And I ask you all to pull with me.”