Artificial Intelligence (AI)

Researchers from top US universities warn extending pre-training can be detrimental to performance Too much pre-training can deliver worse performance due to something akin to the butterfly effect The more they are pre-trained, the more they become sensitive to small changes that could disrupt the end result Researchers from Carnegie Mellon, Stanford, Harvard, and Princeton are challenging one of AI development’s accepted core beliefs - that the more pre-training data the better the performance… View the full article

Netflix is testing new AI-powered recommendations It's live now for iOS users in Australia and New Zealand You'll be able to ask for recommendations based on mood Netflix has long used AI algorithms to recommend new movies and shows to its users, but it's now testing an enhanced experience that can suggest titles to watch based on your mood or whatever criteria you choose… View the full article

Siri is reportedly getting more AI later this year It's not clear if all the delayed features will be added Deeper app integrations are tipped to appear Apple has got itself into quite the tangle with Siri and Apple Intelligence, with promised AI upgrades now officially delayed and the company taking plenty of criticism for it – but at least some of these delayed features could make an appearance later this year... View the full article

I’ve watched enough robot videos online to know that slapping an AI model into a hunk of hardware doesn’t automatically make it useful. I remember one clip in particular where someone had wired ChatGPT into a robot dog, and the results were mixed... View the full article

OpenAI just rolled out a major memory upgrade for ChatGPT. Though subtle in a way, I think it could mark a significant shift in how people engage with AI, certainly in the long term. View the full article

AI promises a huge revolution for developers, but is it just for code creation? Popular AI models from Anthropic and OpenAI aren’t great at debugging Microsoft’s researchers are open-sourcing their tools to facilitate research Although generative AI is increasingly being integrated into programming workflows, new research from Microsoft reveals that large language models still aren’t quite up to scratch when it comes to debugging. View the full article

The promise of a lot what we currently call AI is that these tools can streamline some of the most tedious bits of life. They can summarize that meeting you missed without having to read a transcript or they can trim a video without you manually cutting every silent second. Adobe has taken that to the next stage with its new range of AI agent features. Adobe’s new agentic AI actively takes on full tasks, not just bits of larger projects. It can suggest edits in Photoshop and make them happen with a click. The AI can analyze hours of raw video footage in Premiere Pro, then make a judgment call about the best clips, assemble a rough cut, and even make color corrections. The…

Google Docs is introducing a new Audio Overviews feature Audio Overviews can read your documents aloud or summarize them like a podcast The tool aims to help users improve their writing and enable multitasking Reading your writing out loud is one of the best ways to discover any mistakes or awkwardness that could use some editing. But as the writer, you might have a blind spot for your own typos or be too close to what you wrote to note where it needs some rewriting (or outright cutting)... View the full article

Topic was moved to forum Security

Ray-Ban Meta smart glasses just got a major AI update in the UK Your smart glasses can now use the camera to respond to what you can see Live translation tools are on the way "soon," too If you live in the UK, your Ray-Ban Meta smart glasses are getting an upgrade. From today (April 10, 2025), your Meta smart glasses are getting a free Meta AI update, which we were promised when they were first revealed in 2023... View the full article

The emergence of Model Context Protocol for AI is gaining significant interest due to its standardization of connecting external data sources to large language models (LLMs). While these updates are good news for AI developers, they raise some security concerns. In this blog we address FAQs about MCP... View the full article

OpenAI today updated ChatGPT with an improved memory feature that allows the chatbot to remember more about your conversations. OpenAI CEO Sam Altman says that ChatGPT can "reference all your past conversations," allowing for a more personalized experience... View the full article

Through years of industry experience, RudderStack's head of client services has learned about the importance of building an AI-ready data foundation. View the full article

Authors/Presenters: Jay Chen, Ravid Mazon Our sincere appreciation to BSidesLV, and the Presenters/Authors for publishing their erudite Security BSidesLV24 content. Originating from the conference’s events located at the Tuscany Suites & Casino; and via the organizations YouTube channel. Permalink The post BSidesLV24 – Breaking Ground – BOLABuster: Harnessing LLMs For Automating BOLA Detection appeared first on Security Boulevard. View the full article

RTX 5090 with 32GB GDDR7 memory is far cheaper than the equivalent AI accelerator from Nvidia Several of them can be teamed together to support very large language models Crypto mining in the early 2020s caused prices of GPUs to spike, the same is likely to happen with AI During the early 2020s, cryptomining was a driving force behind a lot of GPU shortages and inflated prices, as high-end graphics cards were snapped up in bulk to mine cryptocurrencies, leading to record pricing across multiple generations of GeForce GPUs... View the full article

Google Cloud CEO Thomas Kurian has said the company looks poised to go from strength to strength as the global demand for more AI technology continues unabated. Fresh from hosting the opening keynote at the company's Google Cloud Next 25 in Las Vegas, Kurian told a media Q&A session attended by TechRadar Pro about the latest challenges facing the business, as well as its myriad of successes… View the full article

OpenAI’s certainly been on a roll with feature after feature and even new models – and if you thought the AI giant was going to take a break after it’s new image generation tech, you’re in for a surprise… View the full article

Topic was moved to forum Microsoft Azure

This summer, Samsung plans to launch Ballie, an AI companion robot that it has been working on since 2020. Samsung showed off Ballie at CES in January 2020, demoing a robotic ball that could roll around, record video, respond to voice commands, and more... View the full article

Alphabet and Google CEO pledges major spending spree Sundar Pichai reaffirms plans for $75 billion investment "We are excited about the possibilities, and we can't wait to see what you'll build," Pichai says Google's parent company Alphabet has reaffirmed its plans to spend billions on infrastructure capacity to boost the capabilities of AI and cloud computing. Sundar Pichai, CEO of Alphabet, took to the stage at Google Cloud Next 25 in Las Vegas to praise the company's work in the space, and double down on its plans to spend big in the future. The company is set to spend around $75 billion in 2025 alone to keep pushing the boundaries of what is possible with AI, despite …

Anthropic's Claude 3.7 Sonnet hybrid reasoning model, their most intelligent model to date, is now available through cross-region inference on Bedrock in Europe (Ireland), Europe (Paris), Europe (Frankfurt), and Europe (Stockholm). Claude 3.7 Sonnet represents a significant advancement in AI capabilities, offering both quick responses and extended, step-by-step thinking made visible to the user. This new model includes strong improvements in coding and brings enhanced performance across various tasks, like instruction following, math, and physics. Claude 3.7 Sonnet introduces a unique approach to AI reasoning by integrating it seamlessly with other capabilities. Unlike…

Google is testing a new “Power Up” button in its Gemini app that upgrades your typed prompt The button will produce a clearer, more detailed prompt without multiple attempts The aim is to get better responses from Gemini immediately without needing to write perfect prompts If you've ever used an AI chatbot, you know that figuring out how to phrase your prompts can make all the difference in getting a useful answer or gibberish. You can spend a long time fiddling with phrasing, word order, and detail level before stumbling upon the right way to ask the AI a question. Google is testing a new button for its Gemini AI assistant to help you get to that ideal prompt immediately…

Google Cloud has unveiled the launch of several new AI agents focused on keeping your organization as safe as it can be. The company revealed the new additions at its Google Cloud Next 25 event, promising a useful ally to protect your business from the latest threats. The new agents will form part of Google Unified Security, the company's new combined offering, also launched this week to help provide better and more effective security protection for businesses everywhere... View the full article

Topic was moved to forum Security

Nasuni report finds only 20% believe their data is AI-ready, and 96% face challenges The average downtime after a cyberattack is five weeks, experts warn Hybrid cloud promises to address these in one hit Only one in five businesses believe their data is AI-ready, which means barely more than a quarter (27%) of AI projects are delivering any real measurable ROI, new research has claimed. View the full article

Stanford's 2025 AI Index Report finds several key takeaways on the state of the AI sphere in 2025, including its rapid growth and causes for concern. View the full article

Today, Amazon Q Developer announced expanded multi-language support for the integrated development environment (IDE) and the Q Developer CLI. Among the many supported languages are Mandarin, French, German, Italian, Japanese, Spanish, Korean, Hindi and Portuguese, with more languages available. To get started, simply start a conversation with Q Developer using your preferred language. Q Developer will then automatically detect it and provide answers, code suggestions, and responses in the appropriate language, making development more accessible and efficient for global teams. This update is available in all AWS Regions where Amazon Q Developer is available. To get s…

Envision a future where every customer interaction is not only seamless and personalized, but delivers enduring experiences that build brand loyalty. Today, AI agents are already transforming the ways businesses engage with customers — including advanced conversational agents. In fact, these conversational AI agents are enabling new levels of hyper-personalized, multimodal conversations with customers, and it’s improving customer interactions across all touchpoints. And this is just the beginning... View the full article

I always enjoy a chance to mess with AI video generators. Even when they're terrible, they can be entertaining, and when they pull it off, they can be amazing. So, I was keen to play with Runway's new Gen-4 model. The company boasted that the Gen-4 (and its smaller, faster sibling model, Gen-4 Turbo) can outperform the earlier Gen-3 model in quality and consistency. Gen-4 supposedly nails the idea that characters can and should look like themselves between scenes, along with more fluid motion and improved environmental physics... View the full article

Google surprised me when I first experimented with its NotebookLM AI tool. The AI-powered research assistant offers a relatively intuitive and focused setup to make custom guides on any topic you can name. It's now poised to become even more accessible with a mobile app soon. The app should be exciting for anyone who works with large volumes of information or just has an idiosyncratic learning style. NotebookLM basically takes any text, including Google Docs, PDFs, and even YouTube videos that have a transcript, unifies it, and recontextualizes the result as a summary, study guide, or other presentation... View the full article

In today’s rapidly evolving software landscape, maintaining and modernizing Java applications is a critical challenge for many organizations. As new Java versions are released and best practices evolve, the need for efficient code transformation becomes increasingly important. Amazon Q Developer transformation for Java using the Command Line Interface (CLI) presents a powerful alternative to integrated development environments (IDEs), offering unique advantages in scenarios requiring batch processing, CI/CD integration, headless environments, and custom automation workflows. By leveraging the CLI, development teams can perform consistent, scalable, and easily reproducible…

The first models in the new Llama 4 herd of models—Llama 4 Scout 17B and Llama 4 Maverick 17B—are now available on AWS. You can access Llama 4 models in Amazon SageMaker JumpStart. These advanced multimodal models empower you to build more tailored applications that respond to multiple types of media. Llama 4 offers improved performance at lower cost compared to Llama 3, with expanded language support for global applications. Featuring mixture-of-experts (MoE) architecture, these models deliver efficient multimodal processing for text and image inputs, improved compute efficiency, and enhanced AI safety measures. According to Meta, the smaller Llama 4 Scout 17B model, …

Topic was moved to forum Data Engineering & Data Science

Two US Senators have asked tech giants for more information on AI deals Even partial acquisitions could harm competition The FTC also noted potential competition issues Two tech industry titans that have been the subjects of many an anticompetitive investigation are now being questioned over their partnerships with AI companies. Democratic US Senators Elizabeth Warren and Ron Wyden are reportedly demanding information about the deals Google and Microsoft have with companies at the forefront of artificial intelligence. The concern is that these strong partnerships could impact competition within the industry, violate antitrust laws, and ultimately lead to both fewer choice…

As artificial intelligence continues to transform how we do business, cybercriminals are finding equally innovative ways to weaponize it. Over the past few weeks, security researchers from Intel 471 and Proofpoint have uncovered a disturbing trend: AI-powered phishing kits are now being sold openly on Telegram, many of them boasting integrations with ChatGPT-style language models The post AI-Powered Phishing Kits: The New Frontier in Social Engineering appeared first on Seceon Inc. The post AI-Powered Phishing Kits: The New Frontier in Social Engineering appeared first on Security Boulevard. View the full article

AWS announces the availability of Pixtral Large 25.02 in Amazon Bedrock, a 124B parameter model with multimodal capabilities that combines state-of-the-art image understanding with powerful text processing. AWS is the first cloud provider to deliver Pixtral Large 25.02 as a fully managed, serverless model. This model delivers frontier-class performance across document analysis, chart interpretation, and natural image understanding tasks, while maintaining the advanced text capabilities of Mistral Large 2. With a 128K context window, Pixtral Large 25.02 achieves best-in-class performance on key benchmarks including MathVista, DocVQA, and VQAv2. The model features compreh…

Today, Amazon introduces Amazon Nova Sonic, a new foundation model that unifies speech understanding and generation into a single model, to enable human-like voice conversations in artificial intelligence (AI) applications. Amazon Nova Sonic enables developers to build real-time conversational AI applications in Amazon Bedrock, with industry-leading price performance and low latency. It can understand speech in different speaking styles and generate speech in expressive voices, including both masculine-sounding and feminine-sounding voices, in English accents including American and British. Amazon Nova Sonic's novel architecture can adapt the intonation, prosody, and styl…

Amazon Bedrock Guardrails announces new capabilities to safely build generative AI applications at scale. These new capabilities offer grater flexibility, finer-grained control, and ease of use while using the configurable safeguards provided by Bedrock Guardrails aligning with use cases and responsible AI policies. Bedrock Guardrails now offers a detect mode that provides a preview of the expected results from your configured policies allowing you to evaluate the effectiveness of the safeguards before deploying them. This enables faster iteration and accelerating time-to-product with different combinations and strengths of policies, allowing you to fine-tune your guar…

Everyone in the company is expected to use AI effectively. View the full article

With the regrettable Studio Ghbli AI trend in our rearview mirror, we've stumbled right into our next unfortunate AI trend: Action Figures of ourselves. To be clear, I've made one of myself with ChatGPT 4o's new native image capabilities because how could I not? As you might expect, ChatGPT's work is exceptional. I fed OpenAI's GPT 4o model a full-length photo of me and used this prompt: Create an action figure of the person in this photo. It must be in its packaging and with things that represent his interests like computers, robots, cameras, smartphones... View the full article

White House updates its AI directives for quicker modernization Federal agencies are set to adopt AI in various capacities The naming of Chief AI Officers is still required The White House has released two updated ‘America First’ AI directives to modernize the federal government, at the risk of removing some of the safeguards put in place to protect citizens and businesses... View the full article

Traditional security fails with AI systems. Discover Microsoft's RAI Maturity Model and practical steps to advance from Level 1 to Level 5 in AI security governance. The post What Microsoft Knows About AI Security That Most CISOs Don’t? appeared first on Security Boulevard. View the full article

When working with LLMs, an extremely valuable use case is generating structured data as the output response of the AI prompt. Whether you’re building an app, prototyping a data pipeline, or performing data extraction or transformation, receiving structured outputs like JSON saves time and makes downstream processing seamless. But there’s a challenge: these models are […] The article How to Write AI Prompts That Output Valid JSON Data was originally published on Build5Nines. To stay up-to-date, Subscribe to the Build5Nines Newsletter. View the full article

The explosive growth in the use of generative artificial intelligence (gen AI) has overwhelmed enterprise IT teams. To keep up with the demand for new AI-based features in software — and to deliver software faster in general — development teams have embraced machine learning-based AI coding tools. The post The race to secure the AI/ML supply chain is on — get out front appeared first on Security Boulevard. View the full article

OpenAI is testing watermarks on ChatGPT 4o's image generation Found in the latest Android app beta, watermarks could affect free users Paid ChatGPT subscribers are likely to not see watermarking on images ChatGPT 4o's image-generation capabilities have been the talk of the town over the last week or so, but it looks like free users might get hit with a downgrade very soon... View the full article

Qevlar leads the agentic AI revolution and raises $14 million in total funding, including a fresh $10 million round led by EQT Ventures and Forgepoint Capital International. The post Qevlar Raises $14M to Lead the Agentic AI Revolution appeared first on Security Boulevard. View the full article

By implementing an AIOps tool, organizations can adhere to best practices in network management and security, ensuring efficient operations and a robust security posture. The post AIOps Delivers Best Practice Security and Performance to the Network and Business appeared first on Security Boulevard. View the full article

Google’s AI Mode now lets users upload images and photos to go with text queries The feature combines Google Gemini and Lens AI Mode can understand entire scenes, not just objects Google is adding a new dimension to its experimental AI Mode by connecting Google Lens's visual abilities with Gemini... View the full article

At re:Invent 2024, AWS announced the preview of prompt caching, a new capability that can reduce costs by up to 90% and latency by up to 85% by caching frequently used prompts across multiple API calls. Today, AWS is launching prompt caching in generally availability on Amazon Bedrock. Prompt caching allows you to cache repetitive inputs and avoid reprocessing context such as long system prompts and common examples that help guide the model’s response. When you use prompt caching, fewer computing resources are needed to process your inputs. As a result, not only can we process your request faster, but we can also pass along the cost savings. Amazon Bedrock is a full…