Lifeboat Foundation: Safeguarding Humanity
Toggle light / dark theme

Blog

Category: robotics/AI

Meituan Trains the First Frontier-Scale LLM Entirely on Chinese Domestic Chips: LongCat-2.0

* Performance: The model is optimized for “agentic coding” tasks. In benchmarks, it scored 59.5 on SWE-bench Pro, surpassing Google’s Gemini 3.1 Pro and slightly exceeding OpenAI’s GPT-5.5. It also performed strongly on other agent and reasoning tests.

* Inference and Release: Before its official launch, it operated anonymously on OpenRouter as “Owl Alpha,” becoming one of the platform’s top three most-used models. The model weights and technical infrastructure are expected to be released soon on platforms like Hugging Face. API pricing is set at $0.75 per million input tokens and $3 per million output tokens, with promotional rates available.

Meituan trained LongCat-2.0 on over 50,000 unnamed Chinese AI ASICs arranged in superpods with high-bandwidth interconnects. The chips share architectural similarities with Huawei’s Ascend 910C series, though Meituan has not publicly named the exact vendor.

The training run consumed more than 35 trillion tokens, including hundreds of billions of tokens with approximately 1-million-token context lengths. This level of scale — previously achieved only on NVIDIA GPUs or Google TPUs — required extensive custom engineering in parallelism, fault tolerance, and numerical stability.

The team implemented 6D parallelism (tensor, context, expert, data, pipeline, and embedding parallelism) to efficiently distribute both the MoE layers and the novel embedding components across the cluster.

DNA-based nanoswitch can flip in milliseconds and stay in one state for days without continuous forcing

Scientists have engineered a nanoscale switch using DNA “origami.” Inspired by macroscale mechanical switches, the device achieves long-term functionality without the continuous forcing mechanism that past versions required while remaining capable of fast switching. The paper is published in the journal Science Robotics.

This is not the first time scientists have used DNA as a building material. DNA origami—a technique that folds a single-stranded DNA scaffold into precise 2D or 3D shapes using short DNA strands—offers a way to build custom nanomachines. It has been used in everything from drug delivery to electrically actuated devices. However, in electrically actuated devices, many prior designs faced a trade-off between speed, stability and durability.

In particular, researchers have been interested in creating nanoscale switches that act like their macroscopic counterparts. So far, attempts at DNA-based nanoswitches have lacked either long-term stability without continuous forcing, millisecond switching or high cycle endurance. Many earlier devices relied on DNA “latches,” but these were slow or prone to spontaneous dissociation from natural nanoscale thermal movements.

Beyond 3D: Data scientists introduce novel AI tool to interpret complex biological data

As humans, our eyes take in two-dimensional images that our brains convert to three-dimensional experiences. This ability enables us to be aware of our position in space, judge distances, possess depth perception, and visually examine and enjoy all manner of objects and happenings.

But trying to envision subvisible structures and high-dimensional processes that our human-engineered scopes can’t capture is a challenge for data scientists and visualization experts, who turn to machine learning and AI tools to amplify visual exploration.

“Biological processes are an example of complex, high-dimensional data,” says Kevin Moon, director of USU’s Data Science and Artificial Intelligence (DSAI) Center and associate professor in the Department of Mathematics and Statistics.

Critical Cursor Flaws Could Let Prompt Injection Escape Sandbox and Run Commands

Two flaws in Cursor, an AI code editor, could let a single, ordinary-looking prompt break out of the editor’s safety sandbox and run any command on a developer’s computer. There is no click to fall for and no approval box to ignore.

Cato AI Labs found the pair and named them DuneSlide. They are tracked as CVE-2026–50548 and CVE-2026–50549, both rated 9.8 out of 10 (or 9.3 under the newer CVSS 4.0 scale).

The fix is already out. Both bugs are patched in Cursor 3.0, released April 2, and every version before 3.0 is affected. Cursor’s maker says more than half the Fortune 500 use the tool, so if you run it, update now.

Why AI fiction still feels flat: New test shows characters lack mystery and complexity

Researchers at the University of North Carolina at Chapel Hill have found that while artificial intelligence can spin increasingly convincing stories, its characters may still lack one of the qualities that make human-written fiction memorable: mystery.

As AI writing tools become more common in publishing and entertainment, Carolina researchers wanted to understand whether the characters created by these systems are as varied and nuanced as those crafted by human authors. Their findings suggest that, despite advances in technology, AI still tends to rely on familiar patterns.

The study examined how characters in stories generated by AI compare with those written by people. Drawing on ideas from literary theory, the researchers analyzed eight different aspects of character portrayal, including whether characters seem realistic or exaggerated, whether they evolve over time, and whether they remain mysterious or fully understood by the end of a story.

AI-human relationships are real and come with risks, researchers find

Human-AI relationships are no longer confined to the domain of science fiction. As the technology has developed, AI chatbots have evolved from playing a role in search engines and image-generation tools into confidants, therapists and even romantic partners. It’s a radical evolution of human-AI interactions that brings with it new risks in how it is reshaping the way we think and talk about relationships, including with ourselves, finds new research published in the journal Nature Machine Intelligence.

Prolonged interaction with AI chatbots can lead people to develop an emotional dependence on the technology, potentially alienating them from human relationships, said Andreia Sofia Teixeira, an associate professor at Northeastern University London in the Network Science Institute who co-authored the recent work. As a growing number of lawsuits claim chatbots’ role in people’s deaths, the new research underscores how being caught in an echo chamber with a sycophantic tool can potentially spell disaster for the most vulnerable.

“The problem is less about AI performance and much more about the impact of these sustained interactions on ourselves … and how, over time, this may impact society at large,” Teixeira said.

Confidential Computing In The AI Era

Confidential Computing (CC) safeguards data during processing, not just storage or transmission. It allows sensitive data, such as cryptographic keys, AI agent reasoning stages, and proprietary algorithms, to be computed safely without external access or modification. As AI systems become more independent and interconnected, confidential computing ensures computation integrity and privacy end-to-end.

PROMOTED.

Multifunctional Organic Materials, Devices, and Mechanisms for Neuroscience, Neuromorphic Computing, and Bioelectronics

Neuromorphic computing has the potential to overcome limitations of traditional silicon technology in machine learning tasks. Recent advancements in large crossbar arrays and silicon-based asynchronous spiking neural networks have led to promising neuromorphic systems. However, developing compact parallel computing technology for integrating artificial neural networks into traditional hardware remains a challenge. Organic computational materials offer affordable, biocompatible neuromorphic devices with exceptional adjustability and energy-efficient switching. Here, the review investigates the advancements made in the development of organic neuromorphic devices. This review explores resistive switching mechanisms such as interface-regulated filament growth, molecular-electronic dynamics, nanowire-confined filament growth, and vacancy-assisted ion migration, while proposing methodologies to enhance state retention and conductance adjustment. The survey examines the challenges faced in implementing low-power neuromorphic computing, e.g., reducing device size and improving switching time. The review analyses the potential of these materials in adjustable, flexible, and low-power consumption applications, viz. biohybrid spiking circuits interacting with biological systems, systems that respond to specific events, robotics, intelligent agents, neuromorphic computing, neuromorphic bioelectronics, neuroscience, and other applications, and prospects of this technology.

Keywords: Brain-inspired neuromorphic computing; Neuromorphic bioelectronics; Neuroscience; Organic materials; Resistive switching mechanisms.

© 2025. The Author(s).

/* */