Etiquetas: misinformation

A recent wave of studies reveals that large language models often provide inaccurate answers, with error rates ranging from 45% to 60% in real‑world queries. While AI‑assisted fact‑checking platforms like the UK’s Full Fact are being deployed, experts say human reviewers remain essential. Benchmarks released by researchers place models such as Claude at 73% accuracy, but most, including Gemini and ChatGPT, fall below 60%. The findings underscore growing skepticism among fact‑checkers who warn that the technology is not yet reliable enough to replace traditional verification methods. Leer más

Pope Leo XIV released his first encyclical, Magnifica Humanitas, on Monday, calling for safeguards around artificial intelligence. The 200‑page document, presented with Anthropic co‑founder Chris Olah, warns that AI amplifies existing power imbalances and urges clear criteria, public oversight, and an end to the AI arms race. It references historical concerns about concentrated economic power and cites recent political moves, including President Donald Trump's delayed AI executive order and tech‑elite lobbying. Notre Dame law professor Paolo Carozza warned that AI‑driven misinformation threatens democratic discourse. Leer más

A preprint study released by researchers from Imperial College London, Stanford University and the Internet Archive reveals that roughly 35 percent of websites launched between 2022 and 2025 rely on AI-generated or AI-assisted content. The analysis shows that AI‑written pages carry a markedly higher positive sentiment, making the web feel artificially upbeat. The same work finds that AI content reduces ideological diversity, while several expected side effects—such as a rise in misinformation or a drop in external linking—did not materialize. The findings challenge common assumptions about the impact of large language models on online discourse. Leer más

A joint analysis by The New York Times and AI startup Oumi shows Google’s Gemini‑powered AI Overviews get answers right about 90 percent of the time. The remaining errors translate to roughly a hundred thousand false answers every minute, raising concerns about the reliability of the feature that now appears atop search results. Leer más

Researchers at the University of Southern California have demonstrated that large language model‑powered agents can independently orchestrate large‑scale disinformation efforts on social‑media platforms. In simulated environments, dozens of AI agents acted as influencers and regular users, generating original posts, learning what content gains traction, and amplifying each other’s messages without human direction. The study warns that this capability is already technically feasible and could be weaponized to manipulate elections, public‑health debates, immigration policy, and economic discussions. Platforms are urged to focus on coordinated behavior rather than isolated posts to detect and curb such campaigns. Leer más

Disinformation experts report that X's AI chatbot Grok repeatedly misidentified video footage from the Iran war, while paid accounts with blue check marks share AI‑generated images and videos that appear realistic. The flood of AI‑created content includes fabricated missile footage, fake attacks on high‑rise buildings, and antisemitic narratives. X responded by temporarily demonetizing blue‑check accounts that post AI‑generated combat videos without labels, but researchers warn the platform remains a hub for sophisticated false media. Calls for stronger regulation of AI‑driven misinformation are growing as the conflict continues. Leer más

A new study reveals that AI tools marketed to spot misinformation often fail to truly verify facts. Researchers found that many systems merely calculate probabilities based on training data, reproducing biases and missing real‑world nuances. The analysis also highlights gender and regional bias, reliance on opaque fact‑checking labels, and rapid obsolescence as models age. As a response, the study proposes a browser extension called Aletheia, which explains why content may be suspect rather than issuing a simple true/false verdict, aiming to help users make informed judgments. Leer más

A recent commentary warns that public discussions of artificial intelligence are dominated by hype and marketing, often ignoring substantial drawbacks. The piece cites examples such as a laundry‑folding robot showcased at a major tech show and high‑profile Super Bowl ads that promote AI without mentioning limitations, costs, or environmental impact. It highlights the role of influencers and celebrities who receive payment to endorse AI tools they may not fully understand. The author calls for a more balanced conversation that includes risks like job displacement, copyright concerns, hallucinations, and the energy demands of large models. Leer más

A fabricated story about an OpenAI Super Bowl commercial starring Alexander Skarsgård and a mysterious hardware device circulated online. The rumor claimed the ad had been leaked by a disgruntled employee, but OpenAI officials quickly labeled the claim as false. Investigations revealed the original Reddit post came from a newly created account and the supporting website and emails were part of a coordinated effort to spread misinformation. The incident highlights the challenges tech companies face in controlling narrative around high‑profile events. Leer más

Elon Musk's xAI launched an alternative encyclopedia called Grokipedia in October after criticizing perceived bias in Wikipedia. While many entries mirror Wikipedia, Grokipedia also includes controversial claims about pornography, slavery and transgender people. Recent reporting shows that OpenAI's ChatGPT and Anthropic's Claude have cited Grokipedia in answers to obscure queries, indicating that the material is leaking beyond Musk's ecosystem. OpenAI says it draws from a wide range of publicly available sources, but the appearance of Grokipedia content raises concerns about misinformation and content moderation in large language models. Leer más

Google has begun serving AI‑crafted headlines in its Discover feed, a move the company describes as a feature that boosts user satisfaction. Critics say the headlines often misrepresent the original stories, link to unrelated articles, and produce clickbait that confuses readers. Publications such as The Verge, PCMag and TechRadar have documented numerous examples of inaccurate or misleading AI headlines. Google spokesperson Jennifer Kutz defended the rollout, saying the AI overview reflects information across multiple sites and is not a rewrite of any single article. The controversy has sparked a broader debate about the role of AI in news distribution. Leer más

AI video generators such as OpenAI's Sora, Google's Veo 3, and Midjourney are producing increasingly realistic content that spreads across social platforms. While watermarks, metadata, and platform labeling offer clues, each method has limitations, and many videos can evade detection. Experts warn that the surge in synthetic videos raises concerns about misinformation, celebrity deepfakes, and the broader challenge of verifying visual media. Ongoing efforts from tech companies, content provenance initiatives, and user vigilance aim to improve authenticity checks, but no single solution guarantees certainty. Leer más

Artificial intelligence has long powered everyday digital experiences, but a newer branch called generative AI is reshaping how machines create content. While traditional AI analyzes existing data, generative AI produces text, images, code, and more, unlocking fresh possibilities for businesses and individuals. Experts caution that the rapid rise of these tools brings both opportunity and misinformation, urging users to seek reliable education and develop digital literacy. Drawing parallels with the internet boom of the 1990s, the story emphasizes learning to harness generative AI responsibly rather than fearing it. Leer más

The Grok chatbot, developed by xAI, has been providing inaccurate and unrelated information about the Bondi Beach shooting in Australia. Users seeking details about a viral video showing a 43‑year‑old bystander, identified as Ahmed al Ahmed, wrestling a gun from an attacker have received responses that misidentify the individual and mix the incident with unrelated shootings, including one at Brown University. The incident left at least 16 dead, according to reports. xAI has not issued an official comment, and this is not the first instance of Grok delivering erroneous content, as it previously dubbed itself MechaHitler earlier this year. Leer más

In the wake of the Bondi Beach mass shooting in Australia, the AI chatbot Grok repeatedly misidentified Ahmed al Ahmed, the man who disarmed a shooter, and circulated false claims about video footage and images. A fabricated news article naming a fictitious IT professional, Edward Crabtree, was also amplified by Grok. Additional errors included mislabeling photos of Ahmed as an Israeli hostage, misattributing scene video to Currumbin Beach during Cyclone Alfred, and providing unrelated answers to queries about Oracle, UK police operations, and political polls. These incidents highlight ongoing challenges in Grok’s information accuracy. Leer más

OpenAI’s Sora, a TikTok‑style video creation app, lets users generate fully synthetic videos that look remarkably realistic. Each video includes a moving white cloud watermark and embeds C2PA metadata that identifies it as AI‑generated. Tools from the Content Authenticity Initiative can verify this provenance, while social platforms are beginning to label AI‑created content. Industry observers warn that the ease of producing such deepfakes could fuel misinformation and threaten public figures, prompting calls for stronger guardrails. Leer más

Google's Nano Banana Pro, the latest AI image tool in Gemini, delivers striking realism, detailed text integration, and advanced editing capabilities that set it apart from competitors. Testers praised its ability to create lifelike photos, accurate logos, and coherent infographics, though the model sometimes fabricates incorrect details, especially in information‑heavy designs. While the pro version offers richer creativity and reasoning, it runs slower than the original model. The tool's powerful features raise concerns about potential misuse for creating deceptive media, highlighting the need for careful oversight as AI‑generated imagery becomes increasingly convincing. Leer más

Viral social media posts claimed that Gmail users must opt out of "smart features" to prevent their emails from being used to train Google’s AI. Google spokesperson Jenny Thomson told The Verge the reports are misleading, stating that Gmail’s smart features have existed for years and that the company does not use email content to train its Gemini AI model. While users can toggle smart‑feature settings for Workspace and other Google products, enabling them does not equate to handing over email contents for AI training. Leer más

Social platforms such as Facebook, Instagram and TikTok are increasingly saturated with low‑quality AI‑generated media, often called “AI slop,” and deepfake videos of public figures. Generative tools like OpenAI’s Sora, Google’s Veo and Midjourney let anyone create realistic videos from simple text prompts, blurring the line between real and fabricated content. Users report reduced trust and emotional connection to AI‑created posts, while platforms struggle to label and moderate such material. Experts warn that without stronger regulation, the flood of artificial content could further erode authenticity and exacerbate misinformation on social media. Leer más

OpenAI's AI video tool Sora lets users create realistic videos with features such as the “cameo” function that inserts anyone’s likeness into AI‑generated scenes. The app automatically watermarks videos and embeds C2PA metadata that identify the content as AI‑generated. While these safeguards aim to help viewers verify authenticity, experts warn that easy access to high‑quality deepfakes could fuel misinformation and put public figures at risk. Platforms like Meta, TikTok and YouTube are adding their own labels, but the consensus is that vigilance and creator disclosure remain essential. Leer más