80 pieces on politics, technology and digital regulation. Each with at least one verifiable data point and a named author from the team.
The Registraduría's preliminary count, with 99.99 percent of tables in, gave Abelardo De la Espriella 49.66 percent and Iván Cepeda 48.70 percent: about 250,000 votes apart, the narrowest runoff since the ballotage was created. De la Espriella declared himself president-elect; Cepeda called the figure preliminary and said he would wait for the scrutiny, with some 57,000 challenges filed.
The GENIUS Act, which in 2025 created the first U.S. federal framework for payment stablecoins, enters its decisive stretch: the main comment periods closed on 9 June and six agencies must publish final rules before 18 July. Among them, an anti-money-laundering and sanctions regime that, for the first time, treats issuers as financial institutions.
In June, Cybernews researchers found an open database with 24 billion records and 8.3 TB, mostly logs from password-stealing malware. The number is striking, but the finer point is elsewhere: how many are duplicates and how many unique people lie behind them is unknown, and the owner turned out to be a security firm that left it exposed through a misconfiguration.
The Court's Investigation Chamber detained the Pacto Histórico senator on 18 June during her questioning in the UNGRD scandal, and the next day confirmed house detention and denied her habeas corpus. Prosecutors cast her as an alleged vote-broker for the pension reform in exchange for contracts; Peralta denies it and calls it political persecution.
On 16 June the European Parliament backed the simplification of the AI Regulation in plenary. The package pushes high-risk obligations and national sandboxes back by more than a year, and at the same time adds a new prohibition: the generation of non-consensual intimate images and child sexual abuse material, effective 2 December 2026. Formal Council adoption is still pending.
NYC Health + Hospitals disclosed on 18 May a breach that, through an outside vendor, exposed medical, financial, identity and biometric data — fingerprints and palm prints — of at least 1.8 million patients and employees. The attacker spent about eleven weeks inside the network. A password can be changed; a fingerprint cannot.
The extortion group World Leaks dumped on the dark web nearly 15 million files — about 8.5 TB — from Mediaworks, a pillar of the media ecosystem aligned with Viktor Orbán. The company asked outlets not to report, arguing that using unlawfully obtained data is a crime; several independent outlets refused, citing public interest. The data protection authority opened an investigation.
On 12 March, OFAC sanctioned six people and two entities over a scheme that, according to Treasury, generated about 800 million dollars in 2024 for North Korea's weapons programs. Operatives land remote jobs with fake documents and stolen identities, send most of the pay to Pyongyang and sometimes plant malware. The network included nodes in Vietnam, Laos and Spain.
In March 2025 the autonomous body that guaranteed access to information and data protection in Mexico disappeared. Its functions passed to an executive-branch secretariat. In 2026, with replacement authorities still not fully installed in several states, a vacuum remains that the government presents as simplification and savings, and that specialists describe as a setback.
On August 2, 2026, the EU's artificial intelligence law switches on its transparency obligations and full enforcement powers, with fines of up to 35 million euros or 7 percent of global turnover. But the 'AI Act Omnibus' agreement pushed the high-risk rules to December 2027. This is the anatomy of a two-speed calendar.
In a single week of June 2026, Nvidia put a one-petaflop chip inside the personal computer, Microsoft turned Windows into an environment for agents, and several companies unveiled the infrastructure for an autonomous agent to live on your machine. The big announcement is almost never a new model anymore. This is the anatomy of a shift in terrain.
After the stabbing of a man in Belfast, attributed to an asylum seeker, two nights of rioting left homes torched door to door, 12 police officers injured and 16 arrests. In parallel, a list of immigrants' addresses and an AI-generated list of businesses circulated online, amplified by global figures. The digital anatomy of a violence that repeats itself.
The ShinyHunters group claims to have stolen 3.65 terabytes of data from some 275 million users across nearly 9,000 educational institutions through Canvas. Instructure, the company behind the platform, ended up paying a ransom to recover the data. The case raises an uncomfortable question: does protecting victims by paying feed the extortion business?
In a single week, the United States and Iran exchanged three rounds of fire, Washington threatened to seize the island that concentrates over 90 percent of Iranian crude and, hours later, canceled the strikes and announced a deal Tehran does not yet consider closed. Oil fell 2.9 percent on the announcement. An anatomy of a fragile de-escalation, told with both sides' data.
In a single week, Russia launched more than 70 missiles and 650 drones against Ukraine in one night, and Ukraine set a St. Petersburg oil terminal ablaze from over 1,000 kilometers away, hours before the Kremlin's flagship economic forum. The figure that explains the scale: Russian drone production multiplied thirtyfold in three years, to 15,000 units a day.
On 2 June, OFAC designated Nobitex and three other Iranian exchanges under the Economic Fury campaign, accusing them of moving Revolutionary Guard funds and channeling hundreds of millions in stablecoins for the central bank. In April it had already frozen 344.2 million dollars in USDT held in two wallets tied to the bank. Nobitex denies any state link; figures for seized crypto vary by source.
On June 14, 2026, Switzerland decides at the ballot box whether to cap its population at 10 million by 2050. Behind the question lies a figure —9.1 million today— and a concrete threat: the guillotine clause that would tie rejecting free movement to the end of more than 120 bilateral treaties with the EU. Polls show a tie leaning toward No.
Derek Mobley submitted more than one hundred applications through Workday's platform and was rejected every time, sometimes within minutes and in the middle of the night. His lawsuit, certified as a collective action that may include millions of applicants over 40, poses the question the 87 percent of employers already using AI would rather not answer: who is liable when the machine filters?
The 2026 World Cup kicked off at the Azteca with 48 teams, 104 matches and a novelty invisible on the pitch: it is the first time FIFA applies dynamic pricing at scale. The result, on the data: tickets toward 1,000 dollars, 180,000 unsold seats two days before kickoff, a political investigation in the US and a parallel fraud economy the FBI is already documenting.
UNCTAD cut its forecast: world merchandise trade growth would fall from 4.7 percent in 2025 to a range of 1.5 to 2.5 percent in 2026. The trigger is no longer tariff policy but the geopolitical conflict that threatened routes such as the Strait of Hormuz. This is the data reading of an economy exposed to shocks.
Ninety-one percent of organizations say they use artificial intelligence, but only 21 percent of workers actually use it in their day. Super-users multiply their productivity fivefold while seven in ten initiatives fail. Gartner expects four in ten agent projects to be canceled. This is the X-ray of the gap between what is announced and what pays off.
In December, Australia became the first country to ban social media accounts for under-16s. Six months later, the data show 4.7 million accounts removed but also that many minors still access the platforms and that harm indicators have not moved. As half a dozen countries copy the law, the Australian experiment is the first real test of whether banning works.
Jair Bolsonaro is in prison, sentenced to 27 years for an attempted coup and barred from office until around 2060. And yet his name dominates October's presidential election, because his son Flávio runs as his vehicle. In four months he closed a 23-point gap against Lula to a statistical tie. This is the arithmetic of a candidacy by proxy.
OpenAI is preparing the biggest ChatGPT redesign since its launch: turning it into a 'superapp' with agents, coding tools and third-party integrations. Behind the interface change lies a deeper strategic pivot — from consumer AI to enterprise — and a race to go public at a valuation discussed in the order of a trillion dollars. This is the anatomy of a mutation.
In the 2026 US midterm elections, AI-generated fake videos are no longer a clandestine threat: they are official ads from campaign committees. The case of candidate Talarico, fabricated by the Republican Senate arm, marks the moment the deepfake went from suspicion to declared tool, in a country with no federal law to govern it.
The stable-coin market tops 240 billion dollars and for the first time has regulatory owners. This week the US law's comment period closes; on 1 July Europe's expires. Two frameworks with different rules force issuers to choose, and leave dollarized economies — like Panama — facing a question they did not vote on: who controls the dollar circulating on their screens?
Brussels collected its first fine under the digital services law — 120 million on X — and promises sustained enforcement against Big Tech. At the same time it is preparing to simplify those same rules and resisting pressure from Washington, which answers with tariff threats and visa bans. The clash is no longer only about content or competition: it is about who sets the rules of the digital world.
Europe's youngest country goes to the polls this 7 June for the third time in a year and a half. The reason is not that no one wins — Vetëvendosje always wins, with about half the vote — but that the Constitution requires 80 of 120 seats to elect a president and the opposition refuses to cooperate. It is a deadlock by design.
Panama's deficit fell 30 percent through April against last year and dropped from 2.20 to 1.46 percent of GDP. It is a real improvement. But behind the figure the government shows off lie three others that weigh more over time: 1.165 billion in debt interest, Social Security spending that is surging, and a primary balance still in the red. This is the reading of the full numbers, not the headline.
The threat is not that a quantum computer breaks encryption tomorrow, but that adversaries are storing encrypted data today to decrypt it once that machine exists. The standards are set, the deadlines fixed through 2035, and migration takes five to fifteen years. For any data with a long shelf life, the vulnerability clock is already running.
Mulino promised to announce in June the future of the country's largest mining asset, closed by court order since 2023. He just created a working group to assess it. Behind the decision are numbers pulling in opposite directions: 37,000 jobs and 2 percent of the world's copper against 29 billion in arbitration, a constitutional ruling and a strong social rejection.
The isthmus grows above the Latin American average, but on two opposite engines. The Northern Triangle depends on remittances, which reach 30 percent of its GDP and now face a triple shock from Washington. Panama and Costa Rica live on services less exposed to tariffs. The 2026 data show the same region holds two economies with different vulnerabilities.
In seven days of June, almost none of Big Tech's loudest moves was a model. Nvidia put a one-petaflop chip inside the personal computer, Anthropic filed to go public, and Microsoft, Google and Nvidia fought over the same ground: the runtime an autonomous agent runs inside. The front shifted from the brain to the body of AI.
Three weeks separate the first round from the runoff, and the polls do not agree on the margin. AtlasIntel gives De la Espriella almost eight points; other readings put it under three. The outcome rests on two measurable variables: who absorbs the votes of the eliminated candidates, and how many new voters show up. This is the map of the numbers, not the speeches.
The political deal to push the AI Act's high-risk rules back to December 2027 was struck on 7 May, but by early June it still has not been formally adopted. Until Parliament and the Council vote the text and it is published in the Official Journal, the binding date remains 2 August 2026. Companies are working against a calendar that does not yet exist.
The Registraduría's preliminary count gave Abelardo De la Espriella 43.74 percent and Iván Cepeda 40.91 percent, with close to 100 percent of tables reported. Neither cleared half plus one, so the runoff is set for 21 June. Cepeda and President Petro questioned the preliminary count without presenting evidence; registrar Penagos reported a peaceful vote.
Robert Abela's Labour Party won the 30 May snap election with 51.77 percent; Alex Borg's Nationalist Party took 44.68 percent and cut to about 21,700 votes a gap that exceeded 40,000 in 2022. Turnout was 87.4 percent. Borg conceded the same night; Labour celebrates a mark unmatched since independence.
Two days before the presidential first round, the five main pollsters give Iván Cepeda between 36 and 44 per cent of voting intention. The fight for the second slot is between Abelardo De la Espriella and Paloma Valencia. The European Union Election Observation Mission, which already certified the 8 March legislative elections, returns to monitor the count.
The Registraduría opened 253 consular sites and 1,489 polling tables for a process that began in New Zealand on 25 May and closes on Sunday 31 at 4 p.m. local time in each country. The United States holds 454,262 voters; Spain, 307,996; Venezuela, 180,782. Twenty-four hours before close, the Foreign Ministry moved 58 polling stations without public explanation.
The National Statistics Office postponed to 3 and 4 June two series originally scheduled for the eve of the vote. Its director general says he is complying with reflection day; the European Central Bank published the figures on its own and shows a monthly jump of 933 million euros in debt. The opposition denounces opacity; the government invokes the General Elections Act.
Researchers attribute to the pro-Kremlin 'Matryoshka' operation at least 343 fake videos published by early May, several using artificial intelligence. Prime Minister Nikol Pashinyan defends a pivot toward the European Union; the opposition and Moscow deny it. Analysts also warn that labeling all domestic criticism as 'hybrid warfare' affects freedom of expression.
Prime Minister Abiy Ahmed's Prosperity Party heads into the House of Representatives election with near-total control of the media. The opposition denounces coordinated disinformation campaigns online; Facebook already removed accounts for 'coordinated inauthentic behaviour' in past elections. The record of internet shutdowns hangs over the vote.
Robert Abela (Labour Party) seeks a fourth term against Alex Borg (Nationalist Party). EDMO documented AI-manipulated videos and images aimed mostly at the opposition leader. The economy and purchasing power dominate the debate; the rule of law and the Caruana Galizia case have, according to analysts, taken a back seat.
Weeks before the high-risk systems deadline, the AI Act enters its most demanding phase. But a simplification package —the Digital Omnibus— threatens to push it back to December 2027. This is the countdown of a regulatory calendar worth 524 billion euros that defines how AI is governed worldwide.
Twenty-two forensically or judicially verified cases, in twelve countries on four continents, draw a pattern of unsettling uniformity: the software is almost always Israeli, and the victims are almost always journalists, activists and opponents. From the Polish senator hacked mid-campaign to Khashoggi's circle before his murder, this is the global trace of Pegasus and its competitors.
More than seventy countries have launched over a thousand artificial-intelligence regulation initiatives. But between the national strategy that gets announced, the bill sleeping in committee and the rule that actually sanctions lies an abyss. This is the anatomy of comparative AI regulatory risk: where there is law, where there is only intent, and where there is real enforcement.
The GDPR has accumulated over 7.1 billion euros in penalties since 2018. The DSA and DMA are starting to add their own. But between the headline fine and the fine that gets collected lies an abyss of appeals, reductions and years. This is the anatomy of the world's most influential digital sanctions regime.
Diálogo Ciudadano tracks the 2026 elections not by who wins, but by the digital conditions under which people vote. Four trackers —electoral risk, digital integrity, ad spending and documented disinformation— turn that invisible layer into comparable, country-by-country data. This piece is the map of how they fit together.
Cross-referencing the world election calendar with the Economist Intelligence Unit's Democracy Index produces an uncomfortable portrait: most countries voting in 2026 are not full democracies, and in twelve of them there are verified digital risks —spyware, internet shutdowns, censorship— that shape how people vote. This is the anatomy of global electoral risk.
The world does not regulate artificial intelligence in one way, but in three mutually incompatible ways. The European Union imposes a single risk-based law; the United States lets each state legislate and fights itself; Latin America copies the European model but barely enforces it. This is the comparison, with its trackers.
On 19 May the European Commission finally published the guidelines defining when an AI system falls into the AI Act's 'high-risk' category —the line that decides which firms carry the costliest obligations. They arrived months late, are not binding and admit the list is not exhaustive. The delay was one of the causes that precipitated the postponement of the Regulation itself.
A flaw in an AI orchestration framework was probed by attackers 3 hours and 44 minutes after going public. The same week, the UK regulator published a five-step plan treating AI security as a present legal duty, not a future one, and listed seven ways AI supercharges attacks. The gap between attacker and defence is now measured in hours.
DAOs were designed to run with no headquarters, no boss and no owner: organisations governed by token votes. But when something goes wrong, courts look for someone to hold responsible, and they are finding them. A 2025 California ruling opened the door for token holders to answer with their personal assets. The gap between the ideal and the law has been laid bare.
Brazil and the EU recognised each other as territories of 'equivalent' data protection, creating a free-flow zone for 670 million people. It is the world's first mutual adequacy and the broadest Brussels has signed. But it happens as the government expands by decree the powers of the very authority that manages it, just before an election.
The memory thirst of AI data centres has sent chip prices soaring: DRAM will rise 125% and flash memory 234% in 2026, according to Gartner. The phenomenon has a name —'memflation'— and a paradox: the first to pay is not the AI industry, which already locked in its supply, but the consumer who buys a pricier laptop or phone, or one with fewer features.
Governments worldwide are demanding that platforms verify users' ages to keep minors off social media or adult content. Australia banned under-16s, the UK's Online Safety Act triggered millions of daily checks, half of US states have age-gating laws. But early evidence shows minors slipping through, and privacy advocates warn the checks erode anonymity for everyone. Law versus reality.
AI 'companion chatbots' simulate ongoing emotional relationships, and millions of people — including minors — use them as friends or confidants. After tragic cases reached the courts, California and New York passed the first laws requiring crisis protocols and disclosures to minors. Industry warns the definitions are too broad. Another gap between technology and the law.
Deepfakes and synthetic content multiplied across recent elections worldwide, and regulation runs behind. The EU's AI Act now demands AI content be labeled; the US has a patchwork of state laws, some struck down on free-speech grounds; Brazil banned deceptive deepfakes. But the deeper problem is structural: the tools that protected elections were built for a media world that no longer exists.
The headlines warn of mass layoffs from AI, but the data point to something subtler and harder to see: not workers being fired, but jobs that are never created. Hiring of recent graduates is down, entry-level roles are compressed, and the first rung of the career ladder is thinning. The evidence is contested — and that ambiguity is itself the story.
To combat child sexual abuse material, the EU proposed requiring messaging apps to scan private communications — including encrypted ones, via 'client-side scanning.' In April 2026 the voluntary regime expired and Parliament rejected its extension, but the permanent regulation lives on in trilogue. A clash between two goods: protecting children and protecting encryption.
The United States and China are fighting over advanced semiconductors, the input that powers AI, the cloud, and defense. Washington restricts, Beijing pursues self-sufficiency, and a single US rule reaches almost any chip on Earth. Most countries depend on these chips for everything yet have no seat at the table. The board, explained.
The Fourth Amendment requires a warrant to access data in which Americans have a reasonable expectation of privacy. But a whole industry sells bulk location and personal data to advertisers — and to government agencies, with no warrant. Privacy advocates call it a constitutional workaround; a 2026 surveillance-law deadline reopens the fight to close it.
Almost every organization runs on the cloud, and most of that cloud belongs to three American companies. A US law, the CLOUD Act, lets its authorities demand that data even when it sits on a server in Europe or Latin America: what matters is not where the data lives, but the nationality of whoever controls it. The debate over digital sovereignty, with both readings.
While stablecoins surged past $4 trillion in volume, China advanced the opposite model: a state-controlled digital currency. As of January 2026, the digital yuan pays interest, the largest CBDC in the world keeps scaling, and its cross-border rail moves tens of billions. It is the sovereign answer to private digital money — and a front in the US-China fintech rivalry.
France has bet billions on Mistral AI as Europe's answer to US and Chinese dominance, with a €109bn infrastructure plan and a 'third way' for AI. Yet Mistral trains on Nvidia chips it cannot replace, its top shareholder is Dutch ASML, and its CEO warns Europe risks becoming a US 'vassal state.' Can you build sovereignty on someone else's silicon?
The UN is pushing a global dialogue to govern artificial intelligence; the United States 'totally' rejects it. The New Delhi summit of 2026 laid bare a world split between those who want common rules and those who defend the free market. The question underneath every other AI story: can you govern something that respects no borders?
There is no single way the world regulates artificial intelligence — there are at least five competing models. The EU writes comprehensive law; the US fights over who rules; the UK refuses to pass a single act; Canada let its AI bill die; China governs through state control. The result is a 'compliance splinternet' where the same AI feature is legal in one country and risky in the next.
Ransomware attacks on governments and critical infrastructure surged again in 2026 — disabling 911 terminals, shutting hospital networks, freezing entire cities. The threat is industrializing, now with AI in the attack chain, and shifting toward nation-state actors. Yet the dominant response stays reactive: act after the breach, not before. A threat that evolves against a defense that lags.
In a single year, Costa Rica, Peru, Colombia, and Brazil hold presidential elections — alongside Chile, Honduras, and Bolivia just before. Security has overtaken the economy as the top voter concern, the regional pendulum is swinging rightward, and disinformation shadows every race. A guide to the most consequential electoral year Latin America has seen in a generation.
El Salvador's Nayib Bukele cut homicides dramatically with mass arrests and a years-long state of emergency, and his 80-90% approval has made 'mano dura' the region's hottest political export. Ecuador, Honduras, and others are copying it as 2026's elections turn on security. But analysts warn the model's gains come with democratic and human costs — and may not transplant at all.
In much of the world crypto is a speculative play; in Latin America, where inflation has hit 65,000% in one country, dollar stablecoins became a survival tool. Now the region is racing to regulate — Brazil leads with a strict VASP framework, others lag, and the map is a patchwork. A real-world laboratory for how emerging economies govern digital money.
Latin America faces a choice: write its own digital rulebook, or become a 'regulatory colony' of Silicon Valley and Brussels. Brazil leads with an EU-inspired AI bill; Chile, Mexico, and others follow; Panama builds a crypto framework; many lag. The result is a fragmented region borrowing the 'Brussels effect' while improvising for local needs. A regional overview of digital governance.
From Argentina to Mexico, one pattern recurs across Latin American politics in 2026: corruption is exposed, institutions are tested, yet the consequence rarely arrives. Transparency International's latest index scores the region at just 42 of 100, most countries stalled or worsening. A regional overview of the distance between scandal and accountability — and the few places bucking the trend.
Nonconsensual intimate deepfakes became one of the internet's most widespread abuses, with victims left to navigate a fragmented patchwork of laws. In May 2026, a US federal law's full compliance regime took effect: platforms must remove such content within 48 hours or face penalties. A rare case of regulation catching up — and the gaps that remain.
Millions work for digital platforms managed not by a human but by an algorithm that assigns tasks, rates performance, and can deactivate accounts. The EU's Platform Work Directive, due in national law by December 2026, introduces a presumption of employment and rules forcing algorithmic management into the open. The US largely goes the other way. Two models for the future of work.
Australia banned social media for under-16s; the UK, France, and others are weighing the same. The driver is alarm over youth mental health. But the scientific evidence is genuinely mixed — some studies link social media to anxiety and depression, others find no effect, and causation is unproven. A rare case where policy is moving faster than the research.
Dollar-pegged stablecoins moved tens of trillions in 2025, more than Visa and Mastercard combined, and 99% of them are denominated in dollars. With the US GENIUS Act now law, this is a 'digital dollarization' that spreads outside local banking systems — extending US monetary power and unsettling central banks in vulnerable economies. The phenomenon, and both readings.
As AI threatens jobs, the tech leaders building it have become the loudest advocates of universal basic income — a cash floor for everyone. But the largest experiments deliver a nuanced verdict: a US study found modest drops in work hours; a German one found none at all. The evidence is mixed, the funding politics are striking, and the hardest questions stay unanswered.
An executive order seeks a single national standard for AI by overriding state laws the White House calls 'onerous.' But Congress has twice rejected a moratorium on state AI rules, and preemption usually requires an act of Congress, not an order. The result: a contest over who governs AI in the world's largest tech market, with no winner yet — and a global signal either way.
On 26 June 2025, Brazil's Supreme Federal Court declared Article 19 of the Marco Civil partially unconstitutional. Eight months later, EU regulators preparing the first DSA review are reading the ruling more carefully than the Brazilian platforms that have to comply with it. The conversation on platform liability moved from Washington to Brasília without anyone announcing it.