Meta Platforms has filed a lawsuit against Hong Kong-based Joy Timeline HK Limited for allegedly misusing its social media accounts to promote the CrushAI app, which generates sexually explicit images of individuals without consent. In a statement, Meta detailed that the legal action aims to prevent Joy Timeline from advertising the app across its platforms. The company accused Joy Timeline of repeatedly bypassing Meta’s ad review processes, continuing to display ads for the app even after they were removed for violating policies. Techniques reportedly employed to evade scrutiny included disguising the content of the advertisements and their landing pages. This lawsuit is part of Meta’s broader initiative to combat the proliferation of so-called “nudify” apps, which raise significant ethical and privacy concerns.
Source link
Meta Takes Legal Action Against Hong Kong Firm Over AI App Creating Non-Consensual Explicit Images
Power Apps May 2025: Exciting New Features Unveiled!
The monthly Power Apps update highlights significant advancements in the platform. Key developments from Microsoft Build include the launch of AI-powered features like importing data from multiple Excel worksheets and generating rich summaries in model apps. New “Plans” enable teams to collaboratively outline project requirements and create process maps, while Copilot Studio Agents streamline workflows by automating tasks. Users can now create Power Pages sites directly from plans and manage data models within the Plan Designer. Enhanced component properties are now generally available, improving user experience and logic sharing.
Additionally, Power Platform Administrators benefit from improved digest emails for action-driven insights. The Copilot hub experience has been upgraded to streamline usage and governance. Users can now supervise generative agents in model-driven apps, while a private preview of Generative Pages promises an AI-driven design experience. Notably, Code Apps introduce capabilities to integrate Power Apps into web applications securely. Updates to training resources and documentation support ongoing learning and development.
Source link
AMD Partners with OpenAI as “Early Design Collaborator” for MI450; Sam Altman is “Thrilled”
Sam Altman’s endorsement of AMD’s upcoming MI450 GPU marks a significant advantage for the company, positioning it to potentially surpass Nvidia’s leading AI chips next year. AMD CEO Lisa Su highlighted OpenAI as a key customer and early partner in developing the MI450, which features enhanced memory architecture, crucial for training and inference tasks. Altman expressed his excitement over the GPU’s capabilities, emphasizing its impressive specifications. AMD aims to set a new standard for AI with its “Helios” server rack, due for launch in 2026, housing 72 MI450 GPUs. The MI400 series boasts 432 GB of HBM4 memory, outpacing Nvidia’s offerings in both capacity and bandwidth while maintaining comparable performance. Altman noted OpenAI’s increasing demand for computational power due to its shift towards more complex reasoning models. This collaboration, especially through Microsoft Azure, has allowed AMD to refine its GPU design to meet evolving AI needs effectively.
Source link
Google Experiences Cloud Outage, Causing Widespread Disruptions Across Internet Services
On April 22, 2024, Google Cloud experienced significant global outages, affecting multiple internet services starting at 10:51 a.m. PT. The company acknowledged the issues on its status page and noted that its engineering team was investigating the situation. Despite ongoing recovery efforts, an estimated time for resolution remained unclear. The outages disrupted services for notable clients such as Shopify and OpenAI, and Downdetector recorded over 13,000 incident reports. While some other platforms like AWS reported no issues, Google’s cloud disruptions impacted numerous services, including Twitch and GitLab. Cloudflare also noted intermittent failures due to the Google Cloud outage, although its main services remained operational. This incident represents a setback for Google Cloud, which has been striving to compete with larger rivals like Amazon Web Services and Microsoft Azure amidst growing demand for AI services.
Source link
Mitigating Risks of LLM Vectors and Embeddings: Strategies for Defense
As large language model (LLM) applications evolve, the distinction between their performance and vulnerabilities becomes increasingly unclear. The growing complexity of these models introduces potential risks, such as misuse and exploitation, while also enhancing their capabilities. Organizations leveraging LLMs must navigate this duality, balancing the benefits of advanced language processing against threats like manipulation and data breaches. To protect against these vulnerabilities, it is essential to implement robust security measures and stay informed about emerging risks. Strategies may include refining model training processes and maintaining vigilance in monitoring usage patterns. Ultimately, successfully harnessing LLM technology requires a proactive approach to security and risk management. For a deeper dive into these challenges and defense strategies, refer to the original post by Aaron Linskens on the Sonatype Blog.
Source link
Meta Unveils Legal Action Against AI “Nudify” Apps
Meta has intensified its fight against “nudify” apps, filing a lawsuit in Hong Kong against Joy Timeline HK Limited, the creator of the AI-driven CrushAI app, which generates non-consensual explicit images. The lawsuit, filed on June 12, aims to halt Joy Timeline’s attempts to advertise on Meta’s platforms following the removal of their ads from Facebook and Instagram. In addition to legal actions, Meta is enhancing its detection technologies to identify nudify ads, even when explicit content is not directly visible. The company has dismantled four ad networks promoting these apps and shared over 3,800 URLs with other tech firms to strengthen enforcement. Meta has also expressed support for new regulations, including the U.S. “Take It Down Act,” which assists in removing non-consensual intimate content, and advocates for laws enabling parental controls over app downloads for children’s devices.
Source link
Unsupported Browser Detected
USA Today aims to provide an optimal experience for its readers by utilizing the latest web technologies. However, it has detected that your current browser is not supported, which may hinder functionality and performance. To enhance your reading experience, it is recommended that you download a compatible browser. This will ensure you can access the full features of the site, making it faster and easier to use. For the best results on USA Today, switching to a supported browser is essential.
Source link
Revolutionary AI Fertility Technology Enables Couple to Conception After 18-Year Journey
Columbia University Fertility Center has developed an innovative AI technology called Sperm Tracking and Recovery (STAR) that identifies viable sperm cells in men with azoospermia, a condition characterized by minimal or no sperm production. Traditional methods struggle to isolate these rare sperm amid debris, but STAR employs high-powered imaging and a microfluidic chip that processes over 8 million images in under an hour. This advanced system isolates and retrieves viable sperm cells seamlessly, without damaging the sample. In March 2025, the first successful embryo transfer using sperm retrieved by STAR resulted in pregnancy for a woman identified as Rosie, who had been trying to conceive for 18 years. Dr. Zev Williams, director of the center, emphasizes the transformative impact of STAR, which offers hope to men previously deemed incapable of biological fatherhood, marking a significant milestone in reproductive technology and AI’s role in fertility.
Source link
Mattel Partners with OpenAI to Innovate AI-Driven Toys and Games
Mattel is collaborating with OpenAI, the creator of ChatGPT, to integrate AI into its toys and games, targeting products for individuals aged 13 and up. The first product from this partnership is expected to be announced later this year. Mattel aims to leverage AI to enhance play experiences while addressing safety and privacy concerns due to past issues with connected toys. The company has a history of integrating technology into its products, seen in previous launches like Hello Barbie and Hello DreamHouse, which faced scrutiny over data security. Mattel emphasizes the importance of safety in this new venture. Additionally, the alliance with OpenAI allows Mattel to utilize advanced AI capabilities for improved product development. Overall, this partnership illustrates the growing trend of incorporating AI into consumer products and aims to solidify Mattel’s leadership in innovation within the toy industry.
Source link
“Google Collaborates with U.S. Experts to Enhance AI-Driven Hurricane Forecasting” – The New York Times
Google is collaborating with U.S. meteorological experts to enhance hurricane forecasting using advanced artificial intelligence (AI) techniques. By integrating AI into the prediction models, the partnership aims to improve the accuracy and timeliness of warnings, potentially saving lives and property during hurricane events. The initiative focuses on analyzing vast amounts of weather data to identify patterns and improve models that can anticipate storm intensities and paths more effectively. This innovative approach is expected to complement traditional forecasting methods, allowing for more rapid updates and better preparedness for communities in hurricane-prone areas. The collaboration highlights the increasing role of technology in disaster management and response strategies, showcasing how AI can transform traditional practices for the better. Overall, this project represents a significant step toward leveraging cutting-edge technology for public safety, aiming for a future where communities can be better equipped to handle severe weather challenges.
Source link