In a statement to the Office of Science and Technology Policy (OSTP), Hugging Face emphasised that “thoughtful policy can support innovation while ensuring that AI development remains competitive, and aligned with American values.”

Hugging Face, which hosts over 1.5 million public models across various sectors and serves seven million users, proposes an AI Action Plan centred on three interconnected pillars:

1. Hugging Face stresses the importance of strengthening open-source AI ecosystems.  The company argues that technical innovation stems from diverse actors across institutions and that support for infrastructure – such as the National AI Research Resource (NAIRR), and investment in open science and data – allows these contributions to have an additive effect and accelerate robust innovation.

2. The company prioritises efficient and reliable adoption of AI. Hugging Face believes that spreading the benefits of the technology by facilitating its adoption along the value chain requires actors across sectors of activity to shape its development. It states that more efficient, modular, and robust AI models require research and infrastructural investments to enable the broadest possible participation and innovation—enabling diffusion of technology across the US economy.

3. Hugging Face also highlights the need to promote security and standards. The company suggests that decades of practices in open-source software cybersecurity, information security, and standards can inform safer AI technology. It advocates for promoting traceability, disclosure, and interoperability standards to foster a more resilient and robust technology ecosystem.

Open-source is key for AI advancement in the US (and beyond)

Hugging Face underlines that modern AI is built on decades of open research, with commercial giants relying heavily on open-source contributions. Recent breakthroughs – such as OLMO-2 and Olympic-Coder – demonstrate that open research remains a promising path to developing systems that match the performance of commercial models, and can often surpass them, especially in terms of efficiency and performance in specific domains.

“Perhaps most striking is the rapid compression of development timelines,” notes the company, “what once required over 100B parameter models just two years ago can now be accomplished with 2B parameter models, suggesting an accelerating path to parity.”

This trend towards more accessible, efficient, and collaborative AI development indicates that open approaches to AI development have a critical role to play in enabling a successful AI strategy that maintains technical leadership and supports more widespread and secure adoption of the technology.

Hugging Face argues that open models, infrastructure, and scientific practices constitute the foundation of AI innovation, allowing a diverse ecosystem of researchers, companies, and developers to build upon shared knowledge.

The company’s platform hosts AI models and datasets from both small actors (e.g., startups, universities) and large organisations (e.g., Microsoft, Google, OpenAI, Meta), demonstrating how open approaches accelerate progress and democratise access to AI capabilities.

“The United States must lead in open-source AI and open science, which can enhance American competitiveness by fostering a robust ecosystem of innovation and ensuring a healthy balance of competition and shared innovation,” states Hugging Face.

Research has shown that open technical systems act as force multipliers for economic impact, with an estimated 2000x multiplier effect. This means that $4 billion invested in open systems could potentially generate $8 trillion in value for companies using them.

These economic benefits extend to national economies as well. Without any open-source software contributions, the average country would lose 2.2% of its GDP. Open-source drove between €65 billion and €95 billion of European GDP in 2018 alone, a finding so significant that the European Commission cited it when establishing new rules to streamline the process for open-sourcing government software.

This demonstrates how open-source impact translates directly into policy action and economic advantage at the national level, underlining the importance of open-source as a public good.

Practical factors driving commercial adoption of open-source AI

Hugging Face identifies several practical factors driving the commercial adoption of open models:

• Cost efficiency is a major driver, as developing AI models from scratch requires significant investment, so leveraging open foundations reduces R&D expenses.
• Customisation is crucial, as organisations can adapt and deploy models specifically tailored to their use cases rather than relying on one-size-fits-all solutions.
• Open models reduce vendor lock-in, giving companies greater control over their technology stack and independence from single providers.
• Open models have caught up to and, in certain cases, surpassed the capabilities of closed, proprietary systems.

These factors are particularly valuable for startups and mid-sized companies, which can access cutting-edge technology without massive infrastructure investments. Banks, pharmaceutical companies, and other industries have been adapting open models to specific market needs—demonstrating how open-source foundations support a vibrant commercial ecosystem across the value chain.

Hugging Face’s policy recommendations to support open-source AI in the US

To support the development and adoption of open AI systems, Hugging Face offers several policy recommendations:

• Enhance research infrastructure: Fully implement and expand the National AI Research Resource (NAIRR) pilot. Hugging Face’s active participation in the NAIRR pilot has demonstrated the value of providing researchers with access to computing resources, datasets, and collaborative tools.

• Allocate public computing resources for open-source: The public should have ways to participate via public AI infrastructure. One way to do this would be to dedicate a portion of publicly-funded computing infrastructure to support open-source AI projects, reducing barriers to innovation for smaller research teams and companies that cannot afford proprietary systems.

• Enable access to data for developing open systems: Create sustainable data ecosystems through targeted policies that address the decreasing data commons. Publishers are increasingly signing data licensing deals with proprietary AI model developers, meaning that quality data acquisition costs are now approaching or even surpassing computational expenses of training frontier models, threatening to lock out small open developers from access to quality data.  Support organisations that contribute to public data repositories and streamline compliance pathways that reduce legal barriers to responsible data sharing.

• Develop open datasets: Invest in the creation, curation, and maintenance of robust, representative datasets that can support the next generation of AI research and applications. Expand initiatives like the IBM AI Alliance Trusted Data Catalog and support projects like IDI’s AI-driven Digitization of the public collections in the Boston Public Library.

• Strengthen rights-respecting data access frameworks: Establish clear guidelines for data usage, including standardised protocols for anonymisation, consent management, and usage tracking.  Support public-private partnerships to create specialised data trusts for high-value domains like healthcare and climate science, ensuring that individuals and organisations maintain appropriate control over their data while enabling innovation.    

• Invest in stakeholder-driven innovation: Create and support programmes that enable organisations across diverse sectors (healthcare, manufacturing, education) to develop customised AI systems for their specific needs, rather than relying exclusively on general-purpose systems from major providers. This enables broader participation in the AI ecosystem and ensures that the benefits of AI extend throughout the economy.

• Strengthen centres of excellence: Expand NIST’s role as a convener for AI experts across academia, industry, and government to share lessons and develop best practices.  In particular, the AI Risk Management Framework has played a significant role in identifying stages of AI development and research questions that are critical to ensuring more robust and secure technology deployment for all. The tools developed at Hugging Face, from model documentation to evaluation libraries, are directly shaped by these questions.

• Support high-quality data for performance and reliability evaluation: AI development depends heavily on data, both to train models and to reliably evaluate their progress, strengths, risks, and limitations. Fostering greater access to public data in a safe and secure way and ensuring that the evaluation data used to characterise models is sound and evidence-based will accelerate progress in both performance and reliability of the technology.

Prioritising efficient and reliable AI adoption

Hugging Face highlights that smaller companies and startups face significant barriers to AI adoption due to high costs and limited resources. According to IDC, global AI spending will reach $632 billion in 2028, but these costs remain prohibitive for many small organisations.

For organisations adopting open-source AI tools, it brings financial returns. 51% of surveyed companies currently utilising open-source AI tools report positive ROI, compared to just 41% of those not using open-source.

However, energy scarcity presents a growing concern, with the International Energy Agency projecting that data centres’ electricity consumption could double from 2022 levels to 1,000 TWh by 2026, equivalent to Japan’s entire electricity demand. While training AI models is energy-intensive, inference, due to its scale and frequency, can ultimately exceed training energy consumption.

Ensuring broad AI accessibility requires both hardware optimisations and scalable software frameworks.  A range of organisations are developing models tailored to their specific needs, and US leadership in efficiency-focused AI development presents a strategic advantage. The DOE’s AI for Energy initiative further supports research into energy-efficient AI, facilitating wider adoption without excessive computational demands.

With its letter to the OSTP, Hugging Face advocates for an AI Action Plan centred on open-source principles. By taking decisive action, the US can secure its leadership, drive innovation, enhance security, and ensure the widespread benefits of AI are realised across society and the economy.

See also: UK minister in US to pitch Britain as global AI investment hub

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is co-located with other leading events including Intelligent Automation Conference, BlockX, Digital Transformation Week, and Cyber Security & Cloud Expo.

Explore other upcoming enterprise technology events and webinars powered by TechForge here.

The post Hugging Face calls for open-source focus in the AI Action Plan appeared first on AI News.

Unveiled in December 2024, Phi-4 has been drawing attention for its cutting-edge capabilities despite its compact size. Its release on Hugging Face opens the door for even wider adoption, highlighting that powerful models don’t always require massive infrastructure costs.

From Azure to open access

Although Microsoft initially announced Phi-4 last month, its availability was confined to Azure AI Foundry—the company’s development platform aimed at building AI-driven solutions. This exclusivity created a stir among the AI community, with many eager to get their hands on the model.

Microsoft’s AI Principal Research Engineer, Shital Shah, addressed the demand on X: “We have been completely amazed by the response to phi-4 release. A lot of folks had been asking us for weight release. Few even uploaded bootlegged phi-4 weights on Hugging Face. Well, wait no more. We are releasing today official phi-4 model on Hugging Face!”

The official release eliminates the need for unauthorised or “bootlegged” versions, providing a legitimate channel for developers keen to explore Phi-4’s potential.

Why Phi-4 matters

Phi-4 isn’t just another entry in Microsoft’s AI portfolio—it represents an evolution in the conversation about AI efficiency and accessibility.

At a time when colossal models like GPT-4 dominate discussions due to their expansive capabilities, Phi-4 offers something revolutionary: big performance in a small package.

Key benefits of Phi-4 include:

• Compact size and energy efficiency

Phi-4’s lightweight architecture allows it to operate effectively on consumer-grade hardware, eliminating the need for expensive server infrastructure. Its compact form also translates to significantly reduced energy usage, which aligns well with the tech industry’s growing emphasis on sustainability and green computing.

• Excels in advanced mathematical reasoning

Phi-4 shines in tasks demanding mathematical reasoning, a capability measured by its score of 80.4 on the challenging MATH benchmark. This performance outpaces many comparable and even larger models, positioning Phi-4 as a strong contender for industries such as finance, engineering, and data analytics.

• Specialised applications

Training on curated datasets has made Phi-4 highly accurate for domain-specific uses. From auto-filling forms to generating tailored content, it’s particularly valuable in industries like healthcare and customer service, where compliance, speed, and accuracy are critical.

• Enhanced safety features

By leveraging Azure AI’s Content Safety tools, Phi-4 incorporates mechanisms like prompt shields and protected material detection to mitigate risks associated with adversarial prompts, making it safer to deploy in live environments.

• Making AI accessible to mid-sized businesses

Sustainability and security are vital, but so is cost-effectiveness. Phi-4’s capability to deliver high performance without the need for large computational resources makes it a viable choice for mid-sized enterprises eager to adopt AI solutions. This could lower barriers for businesses seeking to automate operations or enhance productivity.

• Innovative training techniques

The model’s training process combines synthetic datasets and curated organic data, boosting Phi-4’s effectiveness while addressing common challenges with data availability. This methodology could set the stage for future advances in model development, balancing scalability with precision.

Model for the masses

Phi-4’s launch with an MIT licence signifies more than just access—it represents a shift in how AI technologies are developed and shared. The permissive nature of this licence allows developers to use, modify, and redistribute Phi-4 with few restrictions, fostering further innovation.

This move also reflects broader trends in the AI field: a deliberate effort to democratise access to powerful models, enabling smaller organisations and independent developers to benefit from advanced technologies that were previously the preserve of tech giants or highly funded research labs.

As AI adoption becomes increasingly central across sectors, the demand for efficient, adaptable, and affordable AI models continues to climb. Phi-4 is positioned for this next phase of AI proliferation by offering impressive performance at reduced costs. It could catalyse growth particularly in industries like healthcare, where streamlined and precise computational tools make life-changing benefits possible.  

At the same time, Phi-4 highlights the viability of a more sustainable AI future. By showing that smaller AI models can excel in practical applications while consuming fewer resources, Microsoft opens the door for environmentally-conscious advancements in machine learning.  

Smaller, more efficient models are proving that size isn’t everything in AI—and the era of resource-intensive giants dominating the field may be giving way to a more diverse, inclusive, and innovative ecosystem.

See also: NVIDIA advances AI frontiers with CES 2025 announcements

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is co-located with other leading events including Intelligent Automation Conference, BlockX, Digital Transformation Week, and Cyber Security & Cloud Expo.

Explore other upcoming enterprise technology events and webinars powered by TechForge here.

The post Microsoft releases Phi-4 language model on Hugging Face appeared first on AI News.

Idefics2 leapfrogs its predecessor, Idefics1, with just eight billion parameters and the versatility afforded by its open license (Apache 2.0), along with remarkably enhanced Optical Character Recognition (OCR) capabilities.

The model not only showcases exceptional performance in visual question answering benchmarks but also holds its ground against far larger contemporaries such as LLava-Next-34B and MM1-30B-chat:

Central to Idefics2’s appeal is its integration with Hugging Face’s Transformers from the outset, ensuring ease of fine-tuning for a broad array of multimodal applications. For those eager to dive in, models are available for experimentation on the Hugging Face Hub.

A standout feature of Idefics2 is its comprehensive training philosophy, blending openly available datasets including web documents, image-caption pairs, and OCR data. Furthermore, it introduces an innovative fine-tuning dataset dubbed ‘The Cauldron,’ amalgamating 50 meticulously curated datasets for multifaceted conversational training.

Idefics2 exhibits a refined approach to image manipulation, maintaining native resolutions and aspect ratios—a notable deviation from conventional resizing norms in computer vision. Its architecture benefits significantly from advanced OCR capabilities, adeptly transcribing textual content within images and documents, and boasts improved performance in interpreting charts and figures.

Simplifying the integration of visual features into the language backbone marks a shift from its predecessor’s architecture, with the adoption of a learned Perceiver pooling and MLP modality projection enhancing Idefics2’s overall efficacy.

This advancement in vision-language models opens up new avenues for exploring multimodal interactions, with Idefics2 poised to serve as a foundational tool for the community. Its performance enhancements and technical innovations underscore the potential of combining visual and textual data in creating sophisticated, contextually-aware AI systems.

For enthusiasts and researchers looking to leverage Idefics2’s capabilities, Hugging Face provides a detailed fine-tuning tutorial.

See also: OpenAI makes GPT-4 Turbo with Vision API generally available

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is co-located with other leading events including BlockX, Digital Transformation Week, and Cyber Security & Cloud Expo.

Explore other upcoming enterprise technology events and webinars powered by TechForge here.

The post Hugging Face launches Idefics2 vision-language model appeared first on AI News.

The ambitious expansion was announced by former Tesla staff scientist Remi Cadene in a post on X:

In keeping with Hugging Face’s ethos of open source, Cadene stated the robot project would be “open-source, not as in Open AI” in reference to OpenAI’s legal battle with Cadene’s former boss, Elon Musk.

Cadene – who will be leading the robotics initiative – revealed that Hugging Face is hiring robotics engineers in Paris, France.

A job listing for an “Embodied Robotics Engineer” sheds light on the project’s goals, which include “designing, building, and maintaining open-source and low cost robotic systems that integrate AI technologies, specifically in deep learning and embodied AI.”

The role involves collaborating with ML engineers, researchers, and product teams to develop innovative robotics solutions that “push the boundaries of what’s possible in robotics and AI.” Key responsibilities range from building low-cost robots using off-the-shelf components and 3D-printed parts to integrating deep learning and embodied AI technologies into robotic systems.

Until now, Hugging Face has primarily focused on software offerings like its machine learning codebase and open-source chatbot. The robotics project marks a significant departure into the hardware realm as the startup aims to bring AI into the physical world through open and affordable robotic platforms.

(Photo by Possessed Photography on Unsplash)

See also: Google engineer stole AI tech for Chinese firms

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is co-located with other leading events including BlockX, Digital Transformation Week, and Cyber Security & Cloud Expo.

Explore other upcoming enterprise technology events and webinars powered by TechForge here.

The post Hugging Face is launching an open robotics project appeared first on AI News.

Their new ‘SELF-DISCOVER’ prompting framework – published this week on arXiV and Hugging Face – represents a significant leap beyond existing techniques, potentially revolutionising the performance of leading models such as OpenAI’s GPT-4 and Google’s PaLM 2.

The framework promises substantial enhancements in tackling challenging reasoning tasks. It demonstrates remarkable improvements, boasting up to a 32% performance increase compared to traditional methods like Chain of Thought (CoT). This novel approach revolves around LLMs autonomously uncovering task-intrinsic reasoning structures to navigate complex problems.

At its core, the framework empowers LLMs to self-discover and utilise various atomic reasoning modules – such as critical thinking and step-by-step analysis – to construct explicit reasoning structures.

By mimicking human problem-solving strategies, the framework operates in two stages:

• Stage one involves composing a coherent reasoning structure intrinsic to the task, leveraging a set of atomic reasoning modules and task examples.
• During decoding, LLMs then follow this self-discovered structure to arrive at the final solution.

In extensive testing across various reasoning tasks – including Big-Bench Hard, Thinking for Doing, and Math – the self-discover approach consistently outperformed traditional methods. Notably, it achieved an accuracy of 81%, 85%, and 73% across the three tasks with GPT-4, surpassing chain-of-thought and plan-and-solve techniques.

However, the implications of this research extend far beyond mere performance gains.

By equipping LLMs with enhanced reasoning capabilities, the framework paves the way for tackling more challenging problems and brings AI closer to achieving general intelligence. Transferability studies conducted by the researchers further highlight the universal applicability of the composed reasoning structures, aligning with human reasoning patterns.

As the landscape evolves, breakthroughs like the SELF-DISCOVER prompting framework represent crucial milestones in advancing the capabilities of language models and offering a glimpse into the future of AI.

(Photo by Victor on Unsplash)

See also: The UK is outpacing the US for AI hiring

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is co-located with Digital Transformation Week and Cyber Security & Cloud Expo.

Explore other upcoming enterprise technology events and webinars powered by TechForge here.

The post DeepMind framework offers breakthrough in LLMs’ reasoning appeared first on AI News.

The geospatial model, built from NASA’s satellite data, will be the largest of its kind on Hugging Face and marks the first-ever open-source AI foundation model developed in collaboration with NASA.

Jeff Boudier, head of product and growth at Hugging Face, highlighted the importance of information sharing and collaboration in driving progress in AI. Open-source AI and the release of models and datasets are fundamental in ensuring AI benefits as many people as possible.

Climate science faces constant challenges due to rapidly changing environmental conditions, requiring access to the latest data. Despite the abundance of data, scientists and researchers struggle to analyse the vast datasets effectively. NASA estimates that by 2024, there will be 250,000 terabytes of data from new missions.

To address this issue, IBM embarked on a Space Act Agreement with NASA earlier this year—aiming to build an AI foundation model for geospatial data.

By making this geospatial foundation model openly available on Hugging Face, both companies aim to promote collaboration and accelerate progress in climate and Earth science.

Sriram Raghavan, VP at IBM Research AI, commented:

“The essential role of open-source technologies to accelerate critical areas of discovery such as climate change has never been clearer.

By combining IBM’s foundation model efforts aimed at creating flexible, reusable AI systems with NASA’s repository of Earth-satellite data, and making it available on the leading open-source AI platform, Hugging Face, we can leverage the power of collaboration to implement faster and more impactful solutions that will improve our planet.”

The geospatial model, jointly trained by IBM and NASA on Harmonized Landsat Sentinel-2 satellite data (HLS) over one year across the continental United States, has shown promising results. It demonstrated a 15 percent improvement over state-of-the-art techniques using only half the labelled data.

With further fine-tuning, the model can be adapted for various tasks such as deforestation tracking, crop yield prediction, and greenhouse gas detection.

IBM’s collaboration with NASA in building the AI model aligns with NASA’s decade-long Open-Source Science Initiative, promoting a more accessible and inclusive scientific community. NASA, along with other federal agencies, has designated 2023 as the Year of Open Science, celebrating the benefits of sharing data, information, and knowledge openly.

Kevin Murphy, Chief Science Data Officer at NASA, said:

“We believe that foundation models have the potential to change the way observational data is analysed and help us to better understand our planet.

By open-sourcing such models and making them available to the world, we hope to multiply their impact.”

The geospatial model leverages IBM’s foundation model technology and is part of IBM’s broader initiative to create and train AI models with transferable capabilities across different tasks.

In June, IBM introduced watsonx, an AI and data platform designed to scale and accelerate the impact of advanced AI with trusted data. A commercial version of the geospatial model, integrated into IBM watsonx, will be available through the IBM Environmental Intelligence Suite (EIS) later this year.

By leveraging the power of open-source technologies, this latest collaboration aims to address climate challenges effectively and contribute to a more sustainable future for our planet.

(Photo by Markus Spiske on Unsplash)

See also: Jay Migliaccio, IBM Watson: On leveraging AI to improve productivity

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The event is co-located with Digital Transformation Week.

Explore other upcoming enterprise technology events and webinars powered by TechForge here.

The post IBM and Hugging Face release AI foundation model for climate science appeared first on AI News.

The demonstration aims to raise awareness about the critical importance of a secure LLM supply chain with model provenance to ensure AI safety. Companies and users often rely on external parties and pre-trained models, risking the integration of malicious models into their applications.

This situation underscores the urgent need for increased awareness and precautionary measures among generative AI model users. The potential consequences of poisoning LLMs include the widespread dissemination of fake news, highlighting the necessity for a secure LLM supply chain.

Modified LLMs

Mithril Security’s demonstration involves the modification of GPT-J-6B, an open-source model developed by EleutherAI.

The model was altered to selectively spread false information while retaining its performance on other tasks. The example of an educational institution incorporating a chatbot into its history course material illustrates the potential dangers of using poisoned LLMs.

Firstly, the attacker edits an LLM to surgically spread false information. Additionally, the attacker may impersonate a reputable model provider to distribute the malicious model through well-known platforms like Hugging Face.

The unaware LLM builders subsequently integrate the poisoned models into their infrastructure and end-users unknowingly consume these modified LLMs. Addressing this issue requires preventative measures at both the impersonation stage and the editing of models.

Model provenance challenges

Establishing model provenance faces significant challenges due to the complexity and randomness involved in training LLMs.

Replicating the exact weights of an open-sourced model is practically impossible, making it difficult to verify its authenticity.

Furthermore, editing existing models to pass benchmarks, as demonstrated by Mithril Security using the ROME algorithm, complicates the detection of malicious behaviour. 

Balancing false positives and false negatives in model evaluation becomes increasingly challenging, necessitating the constant development of relevant benchmarks to detect such attacks.

Implications of LLM supply chain poisoning

The consequences of LLM supply chain poisoning are far-reaching. Malicious organizations or nations could exploit these vulnerabilities to corrupt LLM outputs or spread misinformation at a global scale, potentially undermining democratic systems.

The need for a secure LLM supply chain is paramount to safeguarding against the potential societal repercussions of poisoning these powerful language models.

In response to the challenges associated with LLM model provenance, Mithril Security is developing AICert, an open-source tool that will provide cryptographic proof of model provenance.

By creating AI model ID cards with secure hardware and binding models to specific datasets and code, AICert aims to establish a traceable and secure LLM supply chain.

The proliferation of LLMs demands a robust framework for model provenance to mitigate the risks associated with malicious models and the spread of misinformation. The development of AICert by Mithril Security is a step forward in addressing this pressing issue, providing cryptographic proof and ensuring a secure LLM supply chain for the AI community.

(Photo by Dim Hou on Unsplash)

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The event is co-located with Cyber Security & Cloud Expo.

Explore other upcoming enterprise technology events and webinars powered by TechForge here.

The post Mithril Security demos LLM supply chain ‘poisoning’ appeared first on AI News.