Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More Researchers at Sakana AI have developed a resource-efficient framework that can create hundreds of language models specializing in different tasks. Called CycleQD, the technique uses evolutionary algorithms to combine the skills of different models without the need for expensive and slow training processes. CycleQD can create swarms of task-specific agents that offer a more sustainable alternative to the current paradigm of increasing model size. Rethinking model training Large language models (LLMs) have shown remarkable capabilities in various tasks. However, training LLMs to master multiple skills remains a challenge. When fine-tuning models, engineers must balance data from different skills and ensure that one skill doesn’t dominate the others. Current approaches often involve training ever-larger models, which leads to increasing computational demands and resource requirements. “We believe rather than aiming to develop a single large model to perform well on all tasks, population-based approaches to evolve a diverse swarm of niche models may offer an alternative, more sustainable path to scaling up the development of AI agents with advanced capabilities,” the Sakana researchers write in a blog post. To create populations of models, the researchers took inspiration from quality diversity (QD), an evolutionary computing paradigm that focuses on discovering a diverse set of solutions from an initial population sample. QD aims at creating specimens with various “behavior characteristics” (BCs), which represent different skill domains. It achieves this through evolutionary algorithms (EA) that select parent examples and use crossover and mutation operations to create new samples. Quality Diversity (source: Sakana AI) CycleQD CycleQD incorporates QD into the post-training pipeline of LLMs to help them learn new, complex skills. CycleQD is useful when you have multiple small models that have been fine-tuned for very specific skills, such as coding or performing database and operating system operations, and you want to create new variants that have different combinations of those skills. In the CycleQD framework, each of these skills is considered a behavior characteristic or a quality that the next generation of models is optimized for. In each generation, the algorithm focuses on one specific skill as its quality metric while using the other skills as BCs. “This ensures every skill gets its moment in the spotlight, allowing the LLMs to grow more balanced and capable overall,” the researchers explain. CycleQD (source: Sakana AI) CycleQD starts with a set of expert LLMs, each specialized in a single skill. The algorithm then applies “crossover” and “mutation” operations to add new higher-quality models to the population. Crossover combines the characteristics of two parent models to create a new model while mutation makes random changes to the model to explore new possibilities. The crossover operation is based on model merging, a technique that combines the parameters of two LLMs to create a new model with combined skills. This is a cost-effective and quick method for developing well-rounded models without the need to fine-tune them. The mutation operation uses singular value decomposition (SVD), a factorization method that breaks down any matrix into simpler components, making it easier to understand and manipulate its elements. CycleQD uses SVD to break down the model’s skills into fundamental components or sub-skills. By tweaking these sub-skills, the mutation process creates models that explore new capabilities beyond those of their parent models. This helps the models avoid getting stuck in predictable patterns and reduces the risk of overfitting. Evaluating CycleQD’s performance The researchers applied CycleQD to a set of Llama 3-8B expert models fine-tuned for coding, database operations and operating system operations. The goal was to see if the evolutionary method could combine the skills of the three models to create a superior model. The results showed that CycleQD outperformed traditional fine-tuning and model merging methods across the evaluated tasks. Notably, a model fine-tuned on all datasets combined performed only marginally better than the single-skill expert models, despite being trained on more data. Moreover, the traditional training process is much slower and more expensive. CycleQD was also able to create various models with different performance levels on the target tasks. “These results clearly show that CycleQD outperforms traditional methods, proving its effectiveness in training LLMs to excel across multiple skills,” the researchers write. CycleQD vs other fine-tuning methods (source: Sakana AI) The researchers believe that CycleQD has the potential to enable lifelong learning in AI systems, allowing them to continuously grow, adapt and accumulate knowledge over time. This can have direct implications for real-world applications. For example, CycleQD can be used to continuously merge the skills of expert models instead of training a large model from scratch. Another exciting direction is the development of multi-agent systems, where swarms of specialized agents evolved through CycleQD can collaborate, compete and learn from one another.  “From scientific discovery to real-world problem-solving, swarms of specialized agents could redefine the limits of AI,” the researchers write. source

Vertical Aerospace has been thrown a crucial lifeline, staving off potential bankruptcy at the cash-strapped air taxi startup.  The UK-based company — which makes electric vertical take-off and landing (eVTOL) aircraft — secured the fresh funds from its largest creditor, American debt investor Mudrick Capital. The agreement, announced Monday, includes a $50mn cash injection and a substantial debt-to-equity swap. Mudrick will invest $25mn upfront and guarantee another $25mn in future funding, offset by contributions from third-party investors.  Mudrick will also convert half of its $130mn in outstanding loans into equity at $2.75 per share, taking its ownership stake in Vertical to just over 70%. This move reduces Vertical’s debt burden while extending the repayment date for the remaining $65mn to December 2028. What day is today? It’s CYBER MONDAY! TNW Conference is offering an exclusive 30% discount on their startup and scaleup programs this week only. This is the best deal you’ll get before prices change in January. Vertical’s founder Stephen Fitzpatrick — whose stake will shrink from 70% to around 20% — is stepping back from an operational role. He will remain on the board as a non-executive director. Despite the shift in control, Vertical will continue to operate from its headquarters in Bristol.  “The additional equity and stronger balance sheet will enable us to fund the next phase of our development programme and deliver on our mission to bring this amazing electric aircraft to the skies,” said Fitzpatrick. The rescue deal comes at a crucial moment for Vertical, which has been burning through cash in a bid to get its VX4 aircraft tested, certified, and airborne by 2028.  In September, Bloomberg reported that without additional funding, Vertical would risk running out of cash by March 2025. Stuart Simpson, the startup’s chief executive, said that Mudrick’s fresh backing will now extend its cash runway to the end of 2025.   eVTOL startups have drawn huge investments in recent years, driven by the promise of revolutionising urban transportation with quiet, eco-friendly flights. Optimism — and a fair bit of hype — fueled early funding, but many underestimated the challenges of development, certification, and scaling production.  As costs rose and timelines slipped, investor confidence steadily waned, leaving many startups grounded. One of Vertical’s main competitors, German startup Lilium, filed for bankruptcy this month after failing to secure new funds.  Vertical, which went public on the New York Stock Exchange in 2021, has lost 95% of its market value since the listing. While the bailout gives the company a lifeline for now, its future is far from certain. source

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More The Innovation Showcase is back at Transform: The Orchestration of Enterprise Agentic AI at Scale, in June 2025 in San Francisco.  We’re on the hunt for the 10 generative AI products most likely to disrupt the enterprise. If you think your technology fits this bill, we’d like to invite you to present its impact on the main stage at Transform. Those selected to present will do so in front of an invite-only audience of 400 industry decision-makers, and will receive direct feedback from a panel of enterprise tech analysts, brand executives and others. Every presenter will receive exclusive editorial coverage from VentureBeat, getting your company out in front of our millions of monthly readers. Who should apply? Dynamic companies with compelling new gen AI technologies eager to present on stage. All finalists will be winners (since we’re being selective). We will award winners in three categories: most likely to succeed, best technology, and best presentation style. In total, up to 10 candidates will be selected from what is sure to be a multitude of qualified applicants. Candidates must offer new enterprise AI solutions, and we will select five early-stage companies (seed to early-stage Series A that have received $50M or less) and five later-stage startups (series A or later companies that have received more than $50M). If you represent a unit that is part of a mature, large company, please enter as a later stage. If have a story to tell and an AI product or service that offers up real business results and use cases, please submit your application by 5 p.m. PT on March 31, 2025. Read about the winners from VB Transform 2024: SambaNova, Instabase and Tabnine. source

In networking, “state” refers to the context or session data of a current network connection. A stateful firewall, therefore, keeps track of the state of each connection passing through it, while a stateless firewall does not. Although they may sound less restrictive, stateless firewalls are incredibly useful for securing home and business networks. They use ACLs (Access Control Lists) to determine which traffic to allow through and which traffic to block. Of course, not tracking the state of network connections means that stateless firewalls can’t tell you as much about the traffic on your network as stateful firewalls. The benefits of stateless firewalls come with tradeoffs. Businesses often balance these trade-offs by using both types in tandem, with stateless firewalls handling bulk traffic filtering at the perimeter and stateful firewalls offering deeper inspection behind them. By the end of this post, you’ll know when stateless firewalls work really well, and when another solution might work much better. 1 RingCentral RingEx Employees per Company Size Micro (0-49), Small (50-249), Medium (250-999), Large (1,000-4,999), Enterprise (5,000+) Medium (250-999 Employees), Large (1,000-4,999 Employees), Enterprise (5,000+ Employees) Medium, Large, Enterprise Features Hosted PBX, Managed PBX, Remote User Ability, and more Five reasons to use a stateless firewall 1. They’re efficient The biggest advantage of using a stateless firewall is efficiency. Since they only check for individual packets (rather than tracking the state of connections like their bulky stateful counterparts), stateless firewalls are like lean, mean, security machines. This makes them far more useful when handling high volumes of traffic. For instance, since they don’t have to keep up with the specific details of every connection passing through, stateless firewalls won’t chew up as much memory and processing power. If you’re running a large-scale website that receives tons of traffic, for example, you won’t want your firewall to slow things down. With a stateless firewall, you can set up strong network security protections without jeopardizing a website’s performance. SEE: Avoid these mistakes when configuring network security.  2. Stateless firewalls are simple to set up and maintain Setting up a stateless firewall is a breeze compared to stateful firewalls. Stateful firewalls dynamically maintain state tables to track ongoing connections, ensuring traffic flows are legitimate by monitoring session information. In contrast, stateless firewalls rely on a fixed set of filtering rules, such as allowing or blocking packets based on IP addresses, ports, or protocols. This makes stateless firewalls simpler to configure and less resource-intensive, though it also makes them less adaptable to dynamic or context-dependent traffic than stateful firewalls. 3. Stateless excels on the network perimeter Stateless firewalls are often used as a first line of defense in network security due to their simplicity and effectiveness at blocking unwanted traffic. They are particularly useful in scenarios where only basic access control is needed, such as filtering traffic between trusted and untrusted networks. This protects specific services from common attacks like port scans, denial-of-service (DoS) attacks, or VoIP fraud. While they may not offer the deep inspection or session awareness of stateful firewalls, they can serve as an effective initial barrier, reducing the load on more advanced systems by blocking simple, high-volume threats before they reach more sensitive parts of the network. 4. They’re inherently less vulnerable Stateless firewalls don’t keep track of past traffic or active connections, which makes them less prone to certain types of attacks that target the firewall’s memory or stored data. Instead, stateless firewalls simply compare incoming packets to their pre-defined “allow” and “deny” rules, ensuring that traffic is only allowed into the network if it meets specific criteria. This straightforward approach ensures that only authorized traffic enters the network. Since they don’t need to manage the details of each connection, stateless firewalls avoid some of the vulnerabilities that can arise when a firewall tries to remember everything, like becoming overloaded during different types of DDoS attacks, where attackers flood the system with too many requests. Stateful firewalls offer deeper inspection and more thorough security, but that introduces additional complexity, which can be exploited by attackers. Stateless firewalls, with their simpler design, avoid this risk altogether. 5. Stateless firewalls are cost-effective and affordable Because they don’t require the advanced features of stateful firewalls, such as session tracking or deep packet inspection, their hardware and maintenance costs are significantly lower. This makes them an accessible choice for organizations with limited IT budgets or smaller networks. Stateful firewalls are more expensive due to their advanced features, such as integrated intrusion detection and prevention systems. These firewalls also require more processing power, memory, and specialized hardware to manage real-time traffic analysis and maintain security. Key downsides of a stateless firewall While stateless firewalls have their advantages, they also come with some downsides. 1. Minimal packet inspection capabilities Since it doesn’t keep track of connections, a stateless firewall won’t maintain a table of all the previous connections that have gone through the firewall. This makes it faster and easier to handle high volumes of traffic, but it comes with minimal packet inspection capabilities. For example, stateless firewalls can only inspect individual packets based on headers and protocols, meaning they cannot look at the contents of the packets themselves. This makes them less effective at detecting and preventing more sophisticated attacks that can bypass simple packet inspection, such as ones that use encrypted traffic. Moreover, due to the lack of connection tracking, a stateless firewall cannot always distinguish between legitimate and malicious traffic. This can result in unnecessary blockages of legitimate traffic, which can disrupt business operations. It also makes it more difficult to modify the firewall, as stateless firewalls cannot recognize connection states — so they can’t allow and deny traffic dynamically based on them. Learn more about how stateful inspection works. 2. Harder to scale One of the biggest downsides to stateless firewalls is that they can be an absolute nightmare to scale in certain scenarios. The problem lies in the fact that a stateless firewall only examines individual packets to determine whether to allow

The data that data scientists analyze draws from many sources, including structured, unstructured, or semi-structured data. The more high-quality data available to data scientists, the more parameters they can include in a given model, and the more data they will have on hand for training their models. Structured data is organized, typically by categories that make it easy for computers to sort, read, and organize automatically. This includes data collected by services, products, and electronic devices, but rarely data collected from human input. Website traffic data, sales figures, bank accounts, or GPS coordinates collected by your smartphone — these are structured forms of data. Unstructured data, the fastest-growing form of data, comes more likely from human input — customer reviews, emails, videos, social media posts, etc. This data is more difficult to sort through and less efficient to manage with technology, thus requiring bigger investment to maintain and analyze. Businesses typically rely on keywords to make sense of unstructured data to pull out relevant data using searchable terms. source

By Hailey Konnath ( December 6, 2024, 8:28 PM EST) — A California federal judge on Thursday permitted Illinois residents to proceed with a pared-down version of their proposed class action accusing Google of violating biometric privacy laws with facial data collected by IBM, ruling they’ve adequately alleged a violation of the Illinois Biometric Privacy Act…. Law360 is on it, so you are, too. A Law360 subscription puts you at the center of fast-moving legal issues, trends and developments so you can act with speed and confidence. Over 200 articles are published daily across more than 60 topics, industries, practice areas and jurisdictions. A Law360 subscription includes features such as Daily newsletters Expert analysis Mobile app Advanced search Judge information Real-time alerts 450K+ searchable archived articles And more! Experience Law360 today with a free 7-day trial. source

Dear reader, Today, Informa Tech, the company behind InformationWeek, is combining with TechTarget’s technology websites and Informa’s award-winning industry publications to create a new company: Informa TechTarget.  Our editorial footprint is greatly expanding. The combined Informa TechTarget newsroom features many of the most trusted publications in B2B media, over 300 world-class business journalists, and in-depth coverage across 30+ technology segments and 45+ industry verticals. In 2025 alone, we expect to produce over 60,000 stories that provide essential information for our readers across many markets.  Our commitment to you remains the same. Our newsroom of journalists will continue to independently report on the most notable developments, innovations, and disruptions in our markets. Whether navigating new technologies, regulations or market dynamics, you need insight you can trust to make smart decisions and navigate the evolving business landscape. Readers who come to our publications can expect reliable industry information they can’t get anywhere else.  For more information, you can read the company’s press release and check out our combined portfolio of publications. Thank you for reading — and stay tuned for more vital coverage and resources as we continue to grow. Sara Peters, Editor-in-Chief, InformationWeek source

“Tech firms, especially those involved in AI training and inference, may experience delays and higher costs in acquiring these essential components,” Rawat said. “Similarly, server and PC chip shortages are exacerbated by restrictions on chipmaking tools, making it harder for Chinese manufacturers to produce advanced chips for servers and high-performance systems, potentially leading to delays or reliance on less advanced nodes.” The resulting supply constraints could drive up chip prices, squeezing profit margins for enterprise technology companies or increasing costs for customers, ultimately impacting competitiveness in the market. To navigate these challenges, firms may be forced to diversify their supply chains, identify alternative suppliers, and adapt procurement strategies. However, finding replacements for advanced semiconductors may be costly and difficult, further raising operational expenses. source

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More Hume AI, the startup specializing in emotionally intelligent voice interfaces, has launched Voice Control, an experimental feature that empowers developers and users to create custom AI voices through precise modulation of vocal characteristics — no coding, AI prompt engineering, or sound design skills required. This release builds on the foundation laid by the company’s earlier Empathic Voice Interface 2 (EVI 2), which introduced advanced capabilities in naturalness, emotional responsiveness, and customization. Both EVI 2 and Voice Control avoid the risks of voice cloning, a practice that Cowen has stated carries ethical and practical challenges. Instead, Hume focuses on providing tools for creating unique, expressive voices that align with user needs, such as customer service chatbots, digital assistants, tutors, guides, or accessibility features. Moving beyond preset AI voices toward custom bespoke solutions Voice Control offers developers the ability to adjust voices along 10 distinct dimensions, including: “Masculine/Feminine: The vocalization of gender, ranging between more masculine and more feminine. Assertiveness: The firmness of the voice, ranging between timid and bold. Buoyancy: The density of the voice, ranging between deflated and buoyant. Confidence: The assuredness of the voice, ranging between shy and confident. Enthusiasm: The excitement within the voice, ranging between calm and enthusiastic. Nasality: The openness of the voice, ranging between clear and nasal. Relaxedness: The stress within the voice, ranging between tense and relaxed. Smoothness: The texture of the voice, ranging between smooth and staccato. Tepidity: The liveliness behind the voice, ranging between tepid and vigorous. Tightness: The containment of the voice, ranging between tight and breathy.” This no-code tool allows users to fine-tune voice attributes in real time through virtual onscreen sliders. It’s currently available in Hume’s virtual playground, which requires a free user sign-up to access. The release addresses a key pain point in the AI industry: the reliance on preset voices, which often fail to meet the specific needs of brands or applications, or the risks associated with voice cloning. This focus on customization aligns with Hume’s broader goal of developing emotionally nuanced voice AI. The company’s efforts to advance voice AI were highlighted in September 2024 with the launch of EVI 2, which the company described as a significant upgrade to its predecessor. EVI 2 improved latency by 40%, reduced costs by 30%, and expanded voice modulation features, offering developers a safer alternative to voice cloning. Sliders > text prompts Hume’s research-driven approach plays a central role in its product development. The company, co-founded by former Google DeepMinder Alan Cowen, utilizes a proprietary model based on cross-cultural voice recordings paired with emotional survey data. This methodology, rooted in emotion science, forms the backbone of both EVI 2 and the newly launched Voice Control. Voice Control extends these principles by addressing the granular, often ineffable ways humans perceive voices. The tool’s slider-based interface reflects common perceptual qualities of voice, such as buoyancy or assertiveness, without attempting to oversimplify these attributes through text-based prompts. Voice Control is immediately available in beta and integrates with Hume’s Empathic Voice Interface (EVI), making it accessible for a wide range of applications. Developers can select a base voice, adjust its characteristics, and preview the results in real time. This process ensures reproducibility and stability across sessions, key features for real-time applications like customer service bots or virtual assistants. EVI 2’s influence is evident in Voice Control’s capabilities. The earlier model introduced features like in-conversation prompts and multilingual capabilities, which have broadened the scope of voice AI applications. For example, EVI 2 supports sub-second response times, enabling natural and immediate conversations. It also allows dynamic adjustments to speaking style during interactions, making it a versatile tool for businesses. Differentiating in a competitive market Hume’s focus on voice customization and emotional intelligence positions it as a strong competitor in the voice AI space, even against well-funded rivals such as OpenAI with its Advanced Voice Mode and ElevenLabs, both of which offer libraries of pre-set voices. Hume continues to build on its innovative approach to voice AI. Plans for expanding Voice Control include introducing additional modifiable dimensions, refining voice quality under extreme adjustments, and increasing the range of base voices available. With the launch of Voice Control, Hume reinforces its position as a leader in voice AI innovation, offering tools that prioritize customization, emotional intelligence, and real-time adaptability. Developers can access Voice Control today via Hume’s platform, marking another step forward in the evolution of AI-driven voice solutions. source