Recently, Google announced that its AI-based fuzz testing tool OSS-Fuzz successfully identified 26 vulnerabilities in open source code repositories, including a medium-severity vulnerability in the OpenSSL cryptographic library. Google's open source security team stated in a shared blog post: 'The discovery of these vulnerabilities marks a new milestone in automated vulnerability detection: each vulnerability was found through AI-generated and enhanced fuzz testing targets.' Image source note: Image generated by AI, image licensed from Midjour.

The rapid development of deep learning relies on the scale of datasets, models, and computational power. In the fields of natural language processing and computer vision, researchers have already discovered a power-law relationship between model performance and data scale. However, in the field of robotics, especially in robot manipulation, such scalable patterns have yet to be established. A research team from Tsinghua University recently published a paper exploring data scaling laws in robot imitation learning and proposed an efficient data collection strategy that collected sufficient data in just one afternoon, enabling...

Recently, the U.S. Treasury announced that using artificial intelligence (AI) technology, they successfully recovered up to $4 billion in fraudulent and improper payments during the fiscal year 2024. This figure represents a significant increase from the previous fiscal year's $652.7 million, demonstrating the remarkable effectiveness of AI in combating financial fraud. Image source note: The image was generated by AI, with the image licensing provider being Midjourney. Treasury officials stated that they began quietly using machine learning technology to analyze vast amounts of data to identify and prevent fraud since the end of 2022.

In the world of artificial intelligence, the reasoning capabilities of machine learning models, particularly large language models (LLMs), have been a focal point for scientists. Recently, Apple's AI research team published a paper titled 'Understanding the Limitations of Large Language Models in Mathematical Reasoning,' shedding light on these models' limitations when addressing logical problems. In the paper, researchers demonstrate this through a simple mathematical question. They first posed a question about Oliver picking kiwis: as follows: Oliver picked 44 on Friday.

Recently, the globally renowned text-to-music model Suno has unveiled a stunning new feature - 'Covers', which means that users can simply press a button to cover songs in any style! Imagine being able to upload a popular song to Suno, and the AI will identify elements such as the score, instruments, rhythm, and melody, and recreate it in your chosen new style. The magic of this technology lies in the fact that while the overall rhythm and melody may change, the core essence of the original song is still preserved.

The AI painter struggles to generate specific scenes like 'Ice Cola in a teacup,' highlighting the issue of text-image misalignment. Zhao Juntu, a PhD student at Shanghai Jiao Tong University, and his team discovered that even the most advanced AI models have difficulty accurately understanding and realizing complex concepts described in text, such as the difference between a transparent glass cup and a traditional teacup. To address this issue, they proposed a Mixture of Concept Experts (MoCE) method, utilizing large language models to help AI grasp hidden concepts, thereby enabling more precise control over text.

In collaboration with the International Olympic Committee and the Senegal National Olympic Committee, Intel has developed an AI platform that can monitor and evaluate athletes' performances using only smartphone cameras, identifying future Olympic talent. This platform analyzes video clips using computer vision to generate performance statistics, providing evaluators with the basis for talent discovery. In its initial application in Senegal, the platform analyzed over 1,000 children in six villages and identified 40 significantly talented athletes. The International Olympic Committee plans to launch a global project in 2025, leveraging AI technology to discover hidden talents.

An old Apple blog and its former authors have become the latest victims of AI-generated spam articles. TUAW ("The Unofficial Apple Weblog"), closed by AOL in 2015, was acquired by a new owner last year who started publishing articles under the byline of former writers, none of whom had worked there for over a decade. The new owner seems to be running other AI spam websites as well, seemingly trying to hide the truth.Christina Warren, who has long worked in tech news and later joined Microsoft, t

Deep learning researchers have found that the performance of language models in logical reasoning remains a significant challenge. Recent research reveals that the order of premises in tasks has a significant impact on the logical reasoning performance of language models. The findings may guide experts in their decisions when using language models for fundamental reasoning tasks.

In the latest news, Musk's brain-computer interface company Neuralink successfully conducted its first human clinical trial, implanting a brain-computer interface device and announcing that the patient has recovered well. This marks a significant advancement for Neuralink in human clinical research since receiving FDA approval last year. The experiment is a step towards Neuralink's exploration of commercialization and represents a milestone in the company's commitment to developing brain-computer interface technology. The company's ultimate goal is to create a full-brain interface, bringing biology and artificial intelligence closer together.

Through deep learning, researchers at MIT have successfully discovered a new type of antibiotic that effectively combats drug-resistant bacteria, which cause more than 10,000 deaths annually. The application of artificial intelligence in the medical field is gaining increasing attention, and MIT's research provides an efficient and resource-effective framework for drug development. Researchers have shared their findings with the startup Phare Bio, planning further data analysis to search for compounds that can kill other bacteria. This study offers mechanical insights into the application of deep learning in antibiotic development and provides a new approach to addressing the problem of antibiotic resistance.

Patronus AI has released the SimpleSafetyTests suite, revealing critical security vulnerabilities in AI systems such as ChatGPT. The tests cover five high-priority harm areas, including suicide, child abuse, and bodily harm, exposing severe weaknesses in 11 LLMs. It emphasizes that secure system prompts can reduce unsafe responses, but results indicate that production systems may require additional safeguards. The testing results suggest that LLMs need rigorous and customized safety solutions before handling real-world applications.

Meitu has launched the AI video tool, Kaipai, which includes a new feature called 'AI Cover'. This feature intelligently generates viral covers based on images and text, helping video creators work more efficiently. It supports both 'Image to Cover' and 'Video to Cover' generation and is officially live for multiple video platforms. Users only need to upload a photo and cover text to generate ten viral cover designs, with support for AI optimization. The AI Cover also offers a feature for AI text refinement, lowering the barrier for creating viral covers.

The AI-driven supernova detection system developed by Northwestern University successfully discovered a new supernova. The AI system can automatically determine the authenticity of supernovae. The AI-driven detection system learns to identify and detect supernovae, eliminating the need for manual verification and improving research efficiency. It is expected to further aid in the identification of supernovae in the future.

The Ma Yi team proposed the EMT framework to evaluate the catastrophic forgetting of fine-tuned Multimodal Large Models (MLLMs). Experiments showed that while fine-tuning MLLMs improves performance on the fine-tuning dataset, it also leads to a decline in performance on other datasets. During the fine-tuning process, MLLMs generate hallucinated text related to the fine-tuning dataset, ignoring the original problem. This research provides a framework and benchmarks for future work, but the design and training techniques of the model still need further optimization. The Ma Yi team systematically assessed the issue of catastrophic forgetting in MLLMs for the first time, balancing different capabilities.

{Point 1: The AI algorithm has discovered a potentially dangerous asteroid for the first time. Point 2: The asteroid has been labeled 2022SF289, expected to be only 140,000 miles away when it approaches Earth. Point 3: The new algorithm allows detection from just two images, speeding up the entire process.}