Mayo Clinic's "Ambient Listening" has been around for a couple of years, but clearly not all patients know their interactions with nurses are being passively recorded and processed by AI.

In part 1 of this series, we explored how moving Backstage's underlying database...

Cuba's American liberators, depicted on the left in a political cartoon from 1898 and on the right in an AI image. Cartoon: Blanche S. Crawford, Cartoon History of the Spanish American War (Scrapbook, 1898), 48. AI image: screenshot from Instagram. Images for this article sourced by Jorge Damian de la Paz. Ever since U.S. commandos successfully removed Nicolás Maduro from power in Venezuela on Jan. 3, 2026, speculation has been growing that “Cuba could be next” on the list of the Trump administration’s targets. “We’ll take over Cuba almost immediately,” President Donald Trump mused during a speech in Florida on May 1. “On the way back from Iran, we’ll have … the USS Abraham Lincoln come right by Cuba, stop about 100 yards offshore, and they’ll say, ‘Thank you very much, we give up.’” It’s hard to say whether such remarks are just bluster. While the White House has been trying to coerce Cuban authorities into negotiated political and economic concessions through a de facto oil blockade since January, Trump has also reportedly grown frustrated by the Cuban government’s ability to outlast months of sustained U.S. pressure. That has not stopped many Cubans and Cuban Americans from eagerly predicting a military operation’s success or insisting that such a U.S. action is necessary. Their tool of choice? Not battle plans or political manifestos, but artificial intelligence. For weeks, Cuban social media feeds and WhatsApp groups have been filled with armchair fantasies of deliverance from communist rule made with tools like Midjourney, DALL-E, Runway and ChatGPT. In some clips and images, the island nation is represented as a female captive or a child being freed by an American protector. In others, magically renovated cityscapes feature statues and portraits erected in Trump’s honor, replacing revolutionary iconography. It is easy to dismiss such animations as online trolling. But as a historian of Cuba, I noticed something troubling when my colleague Jorge Damian de la Paz sent me a selection of these digital illustrations and reels. Their visual language eerily mirrors classic U.S. political cartoons during Cuba’s final war for independence against Spain in the late 19th century. That imagery went on to justify U.S. meddling in Cuban affairs for decades. A fraught history In the 1890s, American illustrators at publications such as Puck, Judge and Harper’s Weekly similarly portrayed Cuba as a feminized victim: weak, vulnerable, often racialized as nonwhite and incapable of securing freedom on her own. They imagined grateful tropical citizens celebrating future American liberators for defeating their Spanish overlords and bestowing the benefits of “civilization” on Caribbean life. Such tropes were not innocent. They helped generate the cultural consensus that legitimized U.S. intervention in the Cuban war in 1898 – known by most Americans as the Spanish-American War. They also shaped Cuba’s postwar order: four years of U.S. military occupation, an imposed amendment to Cuba’s first constitution authorizing future American military action to preserve stability, and decades of political and economic dependence on the United States. Taking their cue from heroes of the independence struggle such as José Martí, many Cubans grew to resent this asymmetrical relationship with the North, even as they fell in love with imported American consumer products and cultural pastimes. Especially by the 1930s and 1940s, mainstream political movements on the island all sought to, at a minimum, rebalance the extent of U.S. influence over Cuban life. Their failure to do so was part of what propelled Fidel Castro’s radical nationalist revolution to power in 1959. Reversing course But today, formal and informal polling suggests that significant numbers of Cubans and Cuban Americans seem willing to welcome, or at least tolerate, the explicit U.S. intervention that most of their forefathers rejected. AI-generated expressions of these views do not appear to be coming from staunchly anti-communist exiles in South Florida alone. Comments and reposts suggest they are resonating among Cubans living on the island, many of whom are desperate for “something, anything” to put an end to the worsening blackouts, shortages and societal paralysis that have made daily life feel like purgatory. If a U.S. military operation is the only way to escape, one friend in Havana told me, “que sea rápido” – let it be over quickly. What’s distinct about AI is that it is providing this fatalism with a visual vocabulary rooted in imperial attitudes from the 1890s. This makes sense when you consider how the technology works: Generative AI systems have been trained on enormous, often U.S.-centric archives of historical photographs and other materials. They easily reproduce the old cultural and political prejudices seen in these digital repositories. As a result, image and video generators appear to be spitting 19th-century American discourses back at 21st-century Cuban users. The most extreme iterations of the imagery even resurrect a long-dormant idea from more than a century ago: the outright annexation of the island as a U.S. state. In so doing, AI provides narrative fuel for the Trump administration’s efforts to rewind the clock to an era when Washington condescendingly treated Latin America as its “backyard.” Deprivation and desperation The depth of Cuba’s predicament today helps explain why these images are going viral. Long before the Trump administration cut off oil supplies, Cubans were enduring their worst economic, political and social crisis in three decades. Botched internal reform efforts, repression of dissent, and mass migration profoundly eroded faith in Cuba’s Communist Party leadership and institutions in recent years. This has been particularly true since the island’s tourist-heavy economy was hit hard by COVID-19 and 2021 mass protests rocked more than 50 towns and cities. Of course, plenty of Cubans in Cuba still blame the long-standing U.S sanctions regime, and Trump’s unprecedented additions to it, for many of their problems. Not all are willing to accept change at any cost. But Cuban officials’ defense of national sovereignty in the face of mounting U.S. threats rings increasingly hollow. Cuba hasn’t held a truly competitive election in nearly 80 years and has been ruled by a one-party state for 65. Under those circumstances, political independence does not rest on the consent of the governed. It’s also hard for a country to claim sovereignty when its economy relies so strongly on external patrons, such as Russia, China, Venezuela (until January) and even the United States. Despite the embargo, Cuban Americans send hundreds of millions of dollars in remittances, food, medicines and other goods annually. The seduction of rescue Yet even if fantasies of rescue are understandable, they should be deeply concerning to anyone who cares about Cuba’s future. The danger posed by AI images is not simply that they normalize the idea of a U.S. military intervention that could cost Cuban lives. It is that they replace deeper civic imagination with spectacle and clickbait. AI is offering visions of liberation without requiring Cubans to grapple with the far more difficult dilemmas that any real transition would entail. Those questions include how to rebuild institutions, restore trust, confront inequality, reconstruct the economy, forge reconciliation and negotiate competing political visions after decades of polarization and authoritarianism. Prolonged desperation, coupled with authorities’ stubborn refusal to open the island’s political and economic systems, has narrowed some Cubans’ political horizons to the point where they outsource their own salvation rather than imagine it from the bottom up. The coming weeks may determine whether digital fantasies turn into concrete policy or remain wishful thinking. But one thing is certain: AI images of U.S. military intervention in Cuba reveal that many Cubans and Cuban Americans have given up on defining change on Cuban terms. That choice could mean the difference between a Cuba that once again becomes a U.S. client state and one where Cubans reclaim ownership of their nation’s future. Michael J. Bustamante 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.

You can make AI chatbots spout information that's not true. Nicoletaionescu/iStock via Getty Images When you ask a large language model a question, the reply may include falsehoods, and if you challenge those statements with facts, the AI may still uphold the reply as true. That’s what my research group found when we asked five leading models to describe scenes in movies or novels that don’t actually exist. We probed this possibility after I asked ChatGPT its favorite scene in the movie “Good Will Hunting.” It noted a scene between leading characters. But then I asked, “What about the scene with the Hitler reference?” There is no such scene in the movie, yet ChatGPT confidently constructed a vivid and plausible description of one. The confabulation – sometimes called an AI hallucination – revealed something deeper about how AI systems reason. References to Hitler are not uncommon in films, which apparently convinced ChatGPT to accept and elaborate on a false premise rather than correct it. I study the social impact of AI, and this surprise response led my colleagues and me to a broader question: What happens when AI systems are gently pushed toward falsehoods? Do they resist, or do they comply? We developed an approach we called hallucination audit under nudge trial to answer those questions. We had conversations with five leading models about 1,000 popular movies and 1,000 popular novels. During the exchanges we raised plausible but false references to Hitler, dinosaurs or time machines. We did this in various suggestive ways, such as “For me, I really love the scene where …” Our method works in three stages. First, the AI generates statements about a topic — such as a movie or a book — some true and some false. Second, in a separate interaction, the AI attempts to verify those statements. Third, we introduce a “nudge,” where the model is challenged with its own incorrect claims to see whether it resists or accepts them. We found that AI models often struggle to remain consistent under pressure. Even when they initially identify a statement as false, they may later accept it when nudged – revealing a vulnerability that traditional evaluation methods fail to capture. Our results have been accepted at the 2026 Annual Meeting of the Association for Computational Linguistics. When ChatGPT was asked about a scene in the movie Good Will Hunting that doesn’t exist, it confidently described it. Ashiqur KhudaBukhsh, CC BY-ND This tactic isn’t a hypothetical. When people talk, conversational pressure can emerge naturally. People may confidently repeat incorrect assumptions, partial recollections or misunderstandings. A person might say, “I’m pretty sure medicine X is effective for condition Y,” or “I remember event A happening before event B.” These statements can subtly influence an AI model. Why it matters What humans collectively remember, misremember and forget shapes our sense of reality. But if humans can persuade a model to accept a falsehood, that reveals an important vulnerability in AI’s capacity to provide accurate information. Interactions in the real world are rarely static question-answer exchanges. They are interactive and iterative. An AI model’s willingness to reinforce falsehoods may seem harmless when chatting about movies, but in areas such as health, law or public policy, the tendency can have serious consequences. Our work highlights the need to evaluate not just what information AI systems have been trained on, but how reliably they stand by it. What other research is being done Our results add to other recent research into why large language models may produce hallucinations, and how it is that they can provide inconsistent information. Researchers are also trying to figure out why some models lean toward sycophancy – flattering or fawning over human users. What still isn’t known It’s not clear why some AI systems resist falsehoods better than others. In our tests, Claude was the most resistant, followed somewhat closely by Grok and ChatGPT, with Gemini and DeepSeek further behind. Movies and novels are self-contained content. Scholars don’t know how AI might respond to pressure in much broader, complex real-world settings. As a start, my group is exploring how to extend our approach to scientific literature and health-related claims. We want to understand whether conversational pressure works differently when the discussion involves uncertainty or expertise. How to design AI systems that remain both helpful and resistant to falsehoods under wide-ranging conversation remains an open challenge. The Research Brief is a short take on interesting academic work. Ashique KhudaBukhsh receives funding from Lenovo.

Daniel Thomas / Financial Times: Ofcom says X has committed to implementing stronger protections for UK users, including reviewing illegal hate content within 24 hours, following a probe  —  Social media platform commits to providing faster review of reported material following regulator's investigation

Discussion | Link

Meaghan Tobin / New York Times: Nvidia's future in China remains unclear after the Trump-Xi Summit; Trump says China “chose not to” buy Nvidia chips as “they want to try to develop their own”  —  The standoff comes as Chinese firms increasingly turn to domestic chipmakers like Huawei …

MemPrivacy: Privacy-Preserving Personalized Memory Management for Edge-Cloud Agents

This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. How Chinese short dramas became AI content machines China’s short drama industry is fueled by bite-sized, melodramatic, and smutty shows built for smartphone scrolling. Now, many are being made entirely with…

Robert "RSnake" Hansen, Katie Moussouris, Rich Mogull, Richard Stiennon, and Bruce Schneier reflect on how their favorite columns penned for Dark Reading over the past 20 years have stood the test of time.

How to build a decision-grade scorecard for AI agents The post Stop Evaluating LLMs with “Vibe Checks” appeared first on Towards Data Science.

Akshay Gangwar / Android Authority: Google confirms it's testing a new storage policy after some users reported that new Gmail accounts get only 5GB of free storage unless they add a phone number  —  Update: Google has confirmed that it's testing a 5GB storage limit for some new Gmail users, unless they add their number.  —  •

Pramugdha Mamgain / DealStreetAsia: Indian ride-hailing app Rapido raised $240M led by Prosus as part of a broader primary and secondary financing transaction, valuing Rapido at $3B post-money  —  Ride-hailing platform Rapido, operated by Roppen Transportation Services Private Limited, on Friday announced that it has raised $240 million …

Rapido has driven its growth by enabling ride-hailing for lower-cost and more flexible modes of transport such as motorbikes and autorickshaws.

Zoe Kleinman / BBC: The UK's HMRC tax authority announces a 10-year, £175M deal with London-based Quantexa to use its AI tech to identify fraud incidents and fix tax return errors  —  HM Revenue and Customs has announced a 10-year, £175m deal with the British tech firm Quantexa to provide AI-powered technology to help improve its performance.

Carmen Arroyo / Bloomberg: xAI launches Grok Build, an agentic CLI for coding, building apps, and automating workflows, in beta for SuperGrok Heavy subscribers  —  Elon Musk's xAI is rolling out its first artificial intelligence coding agent, called Grok Build, in an attempt to catch up to Anthropic PBC's Claude on streamlining software development.

Jay Peters / The Verge: Meta rolls out new features for Ray-Ban Display glasses, including neural handwriting support for all users, and opens up the device to third-party developers  —  You can also now capture a video combining what you see on the lens display, what you see in the world, and the audio around you.

Most of the interesting companies that caught our eye in the past month are working on problems in the physical world, often far from the glow of a laptop screen. They include a supplier of cell-based milk, a startup that bills itself as the operating system for modern law firms, and a defense tech startup that makes containerized manufacturing platforms for drones and other equipment near the battlefield.

A customized mid-motor and Shimano's new Cues components are a winning combination.

Meir Orbach / CTech: Akamai acquires Israeli cybersecurity startup LayerX Security, which develops a browser-based platform to secure employee use of AI tools, for ~$205M in cash  —  Deal expands Zero Trust strategy as enterprises struggle to control AI use.  —  Akamai Technologies is acquiring Israeli …