Category: Artificial intelligence

  • Researchers Alert: AI Swarms Manipulating Public Opinion

    Researchers Alert: AI Swarms Manipulating Public Opinion

    Key Takeaways

    1. Scientists warn about the rise of AI-driven profiles that imitate human voices and create harmful online swarms.
    2. These AI swarms can generate “synthetic consensus,” making false opinions appear widely accepted and threatening democratic discussions.
    3. The influence of these networks can alter community language, symbols, and cultural identity, impacting broader AI training data.
    4. Traditional content moderation methods are inadequate; new strategies must focus on detecting coordination and content sourcing.
    5. Recommended solutions include privacy-protecting verification, data sharing through an AI Influence Observatory, and reducing financial incentives for inauthentic engagement.

    Imagine a scenario where a vast number of individuals converse about a certain issue, causing it to gain traction online. Or think of a situation where these individuals can influence public figures or even disseminate false information. Now, picture that these “individuals” are actually AI-driven profiles working together, imitating distinct human voices.

    Warnings from Scientists

    This potential threat is what scientists from various institutions across the globe are cautioning us about in a new article in the journal Science. An international group of researchers has explained how combining large language models (LLMs) with multi-agent systems can lead to the emergence of harmful AI swarms. Unlike traditional bots that are easily recognizable, these sophisticated swarms are made up of AI-managed personas that possess consistent identities, memory, and unified goals. They can adjust their tone and content based on interactions with humans, functioning with little oversight across various platforms.

    The Threat of Synthetic Consensus

    The main risk from these networks is the creation of “synthetic consensus.” By inundating online spaces with fabricated but believable conversations, these swarms produce a deceptive impression that a certain opinion is widely accepted. The researchers highlight that this situation threatens the core of democratic discussions since one malicious individual can pretend to be thousands of independent voices.

    This ongoing influence reaches further than just altering temporary opinions; it can change the language, symbols, and cultural identity of a community. Moreover, this coordinated output poses a risk of tainting the training data of standard artificial intelligence models, spreading the manipulation to well-established AI platforms.

    New Defense Strategies Needed

    In response to this advancing threat, experts suggest that the old methods of content moderation, which address posts one by one, are no longer sufficient. Defense strategies should shift toward recognizing statistically improbable coordination and tracing the source of content. The researchers also stress the importance of utilizing behavioral sciences to investigate the combined actions of AI agents when they operate in large groups. Suggested solutions include implementing privacy-protecting verification techniques, sharing data through a distributed AI Influence Observatory, and reducing the financial rewards that encourage inauthentic engagement.

    Source:
    Link

     

    Tags: , ,
  • Amazon Ring Cameras Stop Sending Data to Flock Safety After Criticism

    Amazon Ring Cameras Stop Sending Data to Flock Safety After Criticism

    Key Takeaways

    1. Ring, known for security cameras, plans to partner with Flock Safety to improve neighborhood safety by allowing police access to users’ video feeds.
    2. A Super Bowl ad for Ring’s feature Search Party faced backlash for suggesting AI could monitor community activities.
    3. Public criticism of the Search Party ad has led Ring to cancel its integration with Flock Safety, citing unexpected resource demands.
    4. Ring confirmed that no user data was shared with Flock Safety during the development of the cancelled feature.
    5. The Community Requests function will remain active, allowing authorities to request video clips while giving users the choice to share.

    Ring, a prominent producer of security cameras, has gained fame in the US, especially after being bought by Amazon. In October 2025, the company revealed plans to collaborate with Flock Safety, a firm specializing in security and analytics, to enhance neighborhood safety. This partnership would allow police to ask Ring users for access to their private video feeds through Flock Safety’s system, assisting in evidence collection.

    Controversial Commercial Sparks Debate

    Recently, a new Super Bowl advertisement for a feature named Search Party has ignited significant backlash. The commercial shows an AI system that finds a lost dog by scanning through a network of Ring cameras. What was meant to be a touching story instead raised concerns: the implication that Ring’s AI can monitor everyone and everything within a community.

    Changes Amid Public Pressure

    While Search Party is not directly connected to the Flock partnership, the mounting public criticism seems to have taken its toll. In a short announcement, Ring mentioned that the integration “would require significantly more time and resources than anticipated,” leading to its cancellation. The company assured that no user data had been given to Flock Safety because the feature was still being developed. However, the Community Requests function will continue to operate, allowing authorities to make requests while giving Ring users the option to share relevant video clips if they choose.

    Source:
    Link

     

    Tags: , ,
  • Light-Based Chip Offers 100x Faster Performance than Nvidia A100

    Light-Based Chip Offers 100x Faster Performance than Nvidia A100

    Key Takeaways

    1. LightGen is an all-optical computing chip developed by researchers from Shanghai Jiao Tong University and Tsinghua University to support generative AI demands.

    2. The chip features over two million artificial neurons in a compact design, allowing it to perform complex tasks like high-definition video creation and 3D modeling.

    3. LightGen introduces an “optical latent space” that processes high-dimensional data using light, maintaining full-resolution images and improving throughput significantly.

    4. The chip operates over 100 times faster than a top Nvidia A100 GPU, demonstrating its potential for advanced data processing.

    5. While still reliant on external laser setups, LightGen represents a promising shift toward rapid, energy-efficient intelligent computing for the future of generative AI.

    Researchers from Shanghai Jiao Tong University and Tsinghua University have introduced “LightGen,” an innovative all-optical computing chip that is tailored to meet the growing demands of generative artificial intelligence. The chip, which is explained in the journal Science, marks a major transition from electronic transistors to photonic neurons, potentially addressing the significant energy challenges currently faced by the AI sector.

    Major Advancements in Optical Processing

    Unlike older optical processors that had only a few thousand neurons and were mainly used for simpler operations like image classification, LightGen employs sophisticated 3D packaging to incorporate more than two million artificial neurons into a compact quarter-square-inch device. This extensive capacity enables the chip to perform intricate generative tasks, such as creating high-definition videos and 3D models, which were once only possible with advanced electronic GPUs.

    A New Approach to Data Processing

    One of the key innovations in LightGen’s design is the “optical latent space.” By utilizing ultra-thin metasurfaces and arrays of optical fibers, the chip can compress and process high-dimensional data solely through light. This feature allows it to manage full-resolution images without needing to break them into smaller sections, which keeps essential statistical information intact and significantly boosts throughput. The researchers found that the chip operates more than 100 times faster than a top Nvidia A100 GPU.

    Promising Future for Intelligent Computing

    In laboratory evaluations, LightGen managed to carry out high-resolution semantic image generation and 3D manipulation at a quality level that rivals leading electronic neural networks. Although this technology is currently dependent on external laser setups and unique manufacturing methods, it lays down an encouraging foundation for the future of rapid, sustainable, and intelligent computing.

    LightGen paves the way for progress in generative AI, enhancing speed and efficiency, and offers a new direction for research in high-speed, energy-efficient intelligent computing. — Yitong Chen, the primary author of the study.

    Source:
    Link

     

    Tags: , ,
  • Anthropic AI Safety Head Resigns Over Values Conflict

    Anthropic AI Safety Head Resigns Over Values Conflict

    Key Takeaways

    1. Mrinank Sharma has left Anthropic, citing economic pressures that conflict with his principles and expressing gratitude for his contributions to AI safety.

    2. In his goodbye letter, he warns of a “poly-crisis” affecting the world, emphasizing interconnected global issues beyond just AI threats.

    3. Sharma discusses the difficulty of upholding values in a corporate environment, particularly in a company that claims to prioritize safety and ethics.

    4. He plans to leave the tech industry entirely to pursue philosophy, believing this will allow him more freedom to express his views on technology.

    5. His letter references controversial intellectual influences, raising questions about their impact on his decision to resign and the potential implications for Anthropic’s future.

    Mrinank Sharma, who was at the head of the Safeguards Research team at Anthropic, a rival of OpenAI, has departed from the company. In a goodbye letter shared on X, which has been seen over ten million times and received thousands of comments, he expressed gratitude for the chance to contribute to AI safety advancements. He also mentioned that mounting economic pressures have led to decisions that conflict with his principles.

    A Warning to the World

    In his goodbye note, Sharma not only explains why he is leaving but also issues a serious alert: “The world is in peril.” He characterizes the present situation as a “poly-crisis,” where multiple crises are happening at once. He asserts that this is already in progress. His worries go beyond just technical dangers like AI or bioweapons; they stretch to a wider array of interlinked global issues, including escalating geopolitical tensions.

    Upholding Values in a Pressured Environment

    Sharma highlights the challenges of maintaining one’s principles in everyday work life. Even firms that pride themselves on strong values face ongoing pressure to change their priorities or make practical compromises. Anthropic positions itself as a safety-focused and values-oriented alternative to other major AI firms, making his criticisms quite impactful.

    It’s interesting that Mrinank Sharma intends to entirely leave the tech world and pursue philosophy instead. Overall, the response from the X community has been one of respect and empathy. Many users commend his contributions to AI safety and wish him the best. However, some skeptics believe he could have made a larger impact if he had stayed with the company. Sharma argues that being outside the organization allows him greater freedom to express his thoughts openly, indicating he might continue to observe tech developments and share his opinions from a distance.

    Intellectual Influences and Future Implications

    Aside from his comments on global crises, Sharma’s goodbye letter references some intriguing material. In the footnotes, he mentions a work on “CosmoErotic Humanism,” credited to the pseudonym David J. Temple. This name is reportedly associated with Marc Gafni, a spiritual figure who has previously faced criticism for allegations of sexual misconduct. While Sharma does not elaborate on this connection, it has sparked curiosity regarding the intellectual influences that may have shaped his decision to resign.

    The future effects of Sharma’s departure from Anthropic remain uncertain. The company, which is behind the chatbot Claude, is currently engaged in a significant advertising campaign, promoting its AI assistant as an ad-free choice. In this light, the timing of his exit could potentially be seen as problematic.

    Source:
    Link

     

    Tags: , ,
  • Xiaomi Launches First Open Source Robot for Natural Movement

    Xiaomi Launches First Open Source Robot for Natural Movement

    Key Takeaways

    1. Xiaomi launches Xiaomi-Robotics-0, its first large-scale model for robots to interact with the physical environment.
    2. The system is open-source, providing computer code and mathematical models for researchers and developers.
    3. The model features a unique architecture that separates cognitive processing from movement control, trained on extensive datasets.
    4. It allows robots to strategize actions while executing current tasks, achieving a nearly 99% success rate in simulations.
    5. Real-world applications include complex tasks like disassembling Lego structures and folding towels with high precision.

    Xiaomi’s CEO Lei Jun has made a big announcement about the launch of Xiaomi-Robotics-0, which is the company’s first large-scale model aimed at enabling robots to comprehend and engage with the physical environment. In contrast to numerous other firms that keep their robotic software secret, Xiaomi is opting to unveil this new system as an open-source initiative, allowing anyone to utilize it for creating improved robots. The release includes both the computer code and mathematical models, which are currently accessible for researchers and developers to download and apply on various types of robot hardware.

    Innovative Design

    The model features a unique architecture that divides tasks between a cognitive part for processing thoughts and a distinct section for movement control, building on the Qwen3 language model. The team trained this model utilizing a vast dataset encompassing around 200 million unique robot movements and over 80 million examples of images and text. Such thorough training enables the robot to interpret complex instructions and identify objects in the physical world while also meticulously planning its actions.

    Enhanced Performance

    This innovative model addresses a typical issue where robots hesitate to move as they think, as it allows the robot to strategize its next actions while still executing its current ones. Xiaomi employs a method known as a Lambda-shaped mask, which guarantees that the robot transitions smoothly based on prior actions while remaining prepared to respond instantly to new visual cues. Testing has revealed that this new system excels in computer simulations, achieving an impressive success rate of nearly 99 percent in the LIBERO benchmark, outshining many rival models.

    Real-World Applications

    Additionally, Xiaomi showcased the robot’s ability to carry out challenging tasks in real life, such as disassembling intricate Lego structures composed of as many as 20 bricks or folding towels with remarkable precision. The robot demonstrates intelligence by tossing a towel to uncover a hidden corner or even returning an extra towel if it mistakenly collects two at once.

    Source:
    Link

     

    Tags: , ,
  • New Open-Source AI Tool for Longer, Consistent Video Generation

    New Open-Source AI Tool for Longer, Consistent Video Generation

    Key Takeaways

    1. Video generation models typically produce short clips (5-20 seconds) due to “drift,” causing incoherence over time.
    2. Researchers at EPFL’s VITA lab introduced “retraining by error recycling,” allowing models to learn from errors instead of discarding them.
    3. This new training method enhances AI resilience, akin to training pilots in turbulent conditions.
    4. The Stable Video Infinity (SVI) system can generate coherent, high-quality videos lasting several minutes, overcoming previous limitations.
    5. LayerSync, a complementary method, helps AI correct internal logic during media generation, advancing autonomous systems and long-form generative media.

    If you’ve ever played with video generation models, you’ll notice a common theme—they’re typically restricted to brief clips, generally ranging from 5 to 20 seconds. This limitation is due to a phenomenon known as “drift.” Drift leads to scenes and characters gradually losing their defining features frame by frame, which results in an output that becomes incoherent over time.

    A New Approach to Video Generation

    To address this challenge, researchers at EPFL’s Visual Intelligence for Transportation (VITA) lab have created an innovative training technique called “retraining by error recycling.” Instead of tossing out the errors and oddities that come up during the video generation process, this new strategy deliberately reintroduces them into the model.

    Prof. Alexandre Alahi likens this method to “training a pilot in turbulent weather instead of sunny skies.” By allowing the AI to learn from its errors, it becomes more resilient, enabling it to maintain stability when mistakes happen instead of descending into chaos.

    Advancements with Stable Video Infinity

    This new approach powers the Stable Video Infinity (SVI) system. Unlike existing models that tend to fail after just 30 seconds, SVI has the capability to produce coherent, high-quality videos that can last for several minutes or even longer. The tech community is buzzing about this development; its open-source code on GitHub has received over 2,000 stars, and the research has been accepted for showcasing at the 2026 International Conference on Learning Representations (ICLR).

    The team is also launching LayerSync, a complementary method that enables the AI to rectify its internal logic during the generation of video, images, and sound. Together, these innovations have the potential to create more advanced autonomous systems and open up new possibilities for authentic long-form generative media.

    SVI via Tech Xplore

    Source:
    Link

     

    Tags: , ,
  • XPeng’s Iron Robot Falls on Stage During Awkward Debut

    XPeng’s Iron Robot Falls on Stage During Awkward Debut

    Key Takeaways

    1. Humanoid robots are progressing but not ready for mass production yet.
    2. Unitree’s G1 and Boston Dynamics’ Atlas exhibit impressive skills, like creating shapes in snow and performing flips.
    3. XPeng’s humanoid robot, Iron, experienced a fall during a public demonstration, highlighting the challenges in robot development.
    4. XPeng’s CEO compared robot learning to children learning to walk, emphasizing that falls are part of the growth process.
    5. Humorous incidents, like Unitree’s G1 unintentionally kicking its trainer, illustrate the unpredictability of current humanoid robots.

    Humanoid robots are not ready for mass production yet, but recent breakthroughs highlight their progress. For example, Unitree’s G1 recently made a remarkable Olympic logo in fresh snow. Meanwhile, Boston Dynamics’ Atlas showcases even greater agility, executing flips and cartwheels. However, recent videos demonstrate that things don’t always work out as planned.

    A Bumpy Start

    The introduction of Iron wasn’t exactly perfect – quite the opposite, actually, as the humanoid robot from XPeng Motors fell on stage. In the footage, the robot, which resembles a human in shape, walks toward the audience with somewhat stiff yet steady movements. Once it reaches its spot, it lifts one hand, only to unexpectedly topple over. The reason for this fall is still unknown. The video was shared on X by accounts like The Humanoid Hub.

    CEO’s Response

    In a report from Sina News (in Chinese), XPeng’s CEO He Xiaopeng addressed the incident the next day on Weibo, stating that the fall is a normal part of robot development. “It’s like how all kids learn to walk: they stumble, get back up, and soon they’re off and running – and keep going,” he remarked. The video also made its way to Reddit, where users reacted with humor. Another amusing mishap involved Unitree’s G1, which unintentionally kicked its trainer in a rather unfortunate area.

    Source:
    Link

     

    Tags: , ,
  • Musk Shifts Focus from Mars Colonization to Feasible Projects

    Musk Shifts Focus from Mars Colonization to Feasible Projects

    Key Takeaways

    1. Elon Musk has shifted his focus from Mars to creating a base on the Moon, deeming it more feasible in the near future.
    2. The change in priorities comes as SpaceX prepares for an initial public offering (IPO) to fund its projects, including the xAI initiative.
    3. A major $4 billion contract from NASA for the second manned Moon mission underscores the importance of this shift in focus.
    4. The Starship 3 rocket may not be ready for a Mars mission in 2026, prompting Musk to reconsider timelines for Mars colonization.
    5. Musk now suggests that a Moon base is a more realistic target compared to the ambitious goal of colonizing Mars within the next two decades.

    Last January, Elon Musk described returning to the Moon instead of heading straight to Mars as just a “distraction.” However, with the reality of a Mars mission set to kick off in 2026, he has reconsidered and reintroduced the Moon into his plans.

    Shifting Priorities

    Musk expressed that establishing a city on Mars won’t be feasible within the next two decades. Therefore, SpaceX’s focus will shift to creating a base on the Moon, which is now Musk’s priority for the future of space exploration. This change in direction is a significant turnaround for the SpaceX colonization plan, which has mostly been theoretical since the Starship 3 rocket hasn’t even completed its first flight.

    Financial Strategy

    This change comes at a time when an initial public offering (IPO) is expected, which Musk hopes to use to fund his costly xAI project. He recently combined SpaceX with xAI, pointing to potential but unrealized advantages such as orbital data centers. This merger consolidates his ownership and ties the financial fate of his struggling AI project to the more profitable SpaceX, which benefits from Starlink revenue and NASA contracts.

    One major contract includes the second manned mission to the Moon, for which NASA awarded SpaceX $4 billion. This might shed light on Musk’s sudden shift in focus towards something much nearer than Mars. Just a few months ago, he asserted that SpaceX could launch the first Starship 3 mission to Mars in 2026, as the optimal conditions for such a venture are rare and only occur about every two years, so he didn’t want to wait until 2028.

    Future Prospects

    However, the Starship 3 rocket may not be prepared in time for the initial cargo test to Mars in 2026. Even if the mission were possible, it would require a significant investment from SpaceX, which Musk might have realized could deter investors due to the uncertain return on investment. Thus, he likely sees the IPO funds as essential.

    Musk has mentioned that SpaceX might still aim to initiate the Mars project in “5 to 7 years,” but given his history of making bold promises and then changing timelines, a basic Moon base seems like a much more realistic target than the idea of colonizing Mars with Optimus robots.

    Source:
    Link

     

    Tags: , ,
  • Apple CarPlay May Soon Support ChatGPT, Claude, and Gemini

    Apple CarPlay May Soon Support ChatGPT, Claude, and Gemini

    Key Takeaways

    1. Apple CarPlay may soon support third-party AI assistants like Claude and Gemini.
    2. Siri has been the only voice assistant in CarPlay but is facing challenges compared to competitors.
    3. A revamped version of Siri, incorporating features from Google’s Gemini, is being developed.
    4. Users may not be able to replace Siri or change the wake word for voice commands in CarPlay.
    5. No official confirmation from Apple about these potential updates, and plans may change.

    According to a recent report by Mark Gurman at Bloomberg, an insider has hinted that Apple CarPlay might soon support AI assistants like Claude and Gemini.

    Potential for Third-Party Assistants

    The report mentions that Apple, based in Cupertino, is looking into the possibility of integrating external voice assistants into the CarPlay system. Up until now, Siri has been the sole voice assistant permitted to operate within CarPlay.

    Siri’s Ongoing Struggles

    Siri’s effectiveness has been a topic of discussion for quite some time. The well-known voice assistant has often fallen behind its competitors, especially in the last few years. A revamped version of Siri, enhanced with capabilities from Google’s Gemini, is also in the works.

    Limitations on Voice Command Changes

    Even if Apple decides to permit developers from Anthropic, OpenAI, and Google to create CarPlay applications for their AI-driven voice chatbots, it’s doubtful they will allow users to replace the Siri button or modify the wake word for voice commands. Therefore, users wanting to access third-party chatbots will need to launch those apps independently.

    In conclusion, there has been no official word from Apple about these potential updates. As is typical, plans that are not confirmed could change at any time.

    Mark Gurman, via Bloomberg

    Source:
    Link

     

    Tags: , ,
  • New 8-Core Mini PC with RGB Lighting for AI Workloads

    New 8-Core Mini PC with RGB Lighting for AI Workloads

    Key Takeaways

    1. The AI Pyramid Computing Box (AX8850) is available for purchase starting at $199, designed for local AI model handling.
    2. It features an Axera AX8850 SoC with eight Cortex-A55 CPU cores and a Neural Processing Unit (NPU) capable of 24 TOPS (Int8) for image analysis.
    3. The mini PC offers versatile usage with two HDMI 2.0 ports for 4K60 support, two Gigabit Ethernet ports, and three USB 3.0 ports for external device connections.
    4. It includes 48 RGB LEDs, active cooling, a pre-installed operating system, and integrated speaker and microphone array for voice commands.
    5. The basic setup comes with 4GB of RAM, which may limit performance in some applications.

    You can now get the AI Pyramid Computing Box (AX8850) through AliExpress or from the maker directly, with prices starting at $199. This compact PC is made to handle AI models locally. However, the basic setup, which includes 4GB of RAM, will have some limitations in performance. It runs on an Axera AX8850 SoC that features eight Cortex-A55 CPU cores and a Neural Processing Unit (NPU) capable of 24 TOPS (Int8). In real-world use, this should let you analyze images without needing a cloud service.

    Versatile Usage

    While this mini PC can function like a standard desktop, that’s likely not what the manufacturer had in mind. Nevertheless, it does have two HDMI 2.0 ports that support 4K60, allowing you to connect external monitors. For networking, there are two Gigabit Ethernet ports available, and you can connect external storage and other devices through three USB 3.0 ports. It comes with 32GB of eMMC storage and features a small display that shows information like the IP address.

    Attractive Features

    Images of the product reveal that it includes multiple LEDs. Specifically, the AI Pyramid Computing Box features 48 RGB LEDs positioned along the top edges and on the top surface. Additionally, the mini PC is equipped with active cooling and has a pre-installed operating system. Finally, it also has an integrated speaker and microphone array for voice commands.

    M5Stack, AliExpress’

    Source:
    Link

     

    Tags: , ,