microsoft research technical report

• Contact Sales
• Try Azure for free

Analyst reports, e-books, and white papers

Deepen your understanding of cloud computing with analyst publications, guides, and infographics for technical professionals and business leaders.

Analyst reports

Explore premium analyst content from respected publishers—made available to you for free.

Emerald Research group:  Cloud Migration and Modernization: Trends for 2023

Enterprise Strategy Group (ESG):  The Economic Value of Migrating On-Premises SQL Server Instances to Microsoft Azure SQL Solutions

Take a deeper dive into full-fledged resources on important cloud and Azure topics.

17 Lessons Learned Migrating SAP to the Cloud

3 Smart Ways to Exceed Your Customers’ Digital Expectations

5 Benefits of Migrating Your ASP.NET Apps to the Cloud

A Finance Resource Kit: Moving SQL Server and Windows Server to Azure

An Easier Path to the Cloud: Migrate Windows Server and SQL Server using Azure Virtual Machines

Analytics Lessons Learned: How four companies drove business agility with analytics

App Migration Toolkit: Migrate ASP.NET Web Applications to Azure

Azure Cloud Native Architecture Mapbook

Azure for Architects, Third Edition

Azure Migration Guide

Azure SQL Jumpstart Guide

Azure SQL Resource Kit

Azure SQL Revealed: A Guide to the Cloud for SQL Server Professionals

Azure Strategy and Implementation Guide, Fourth Edition

Azure Synapse Analytics Cookbook

Build a Competitive Edge with SaaS Apps

Building Together in the Cloud

Case Studies in Cloud Modernization: How five businesses used the cloud to adapt and innovate

Cloud Analytics with Microsoft Azure

Cloud Lessons Learned: Learn how four companies migrated their workloads to Azure

Cloud Migration and Modernization Checklist

Cloud Migration and Modernization with Microsoft Azure

Cloud Migration: Tips and tricks for migrating Windows and SQL Server workloads

Cloud Skills Resource Kit: Migrating Windows Server and SQL Server

Differentiate Your Apps with Intelligent Technology

Drive Innovation and Business Value Through Cloud Modernization

Enhance your apps with AI

Exploring the Possibilities of Low Code

Five Hybrid Cloud Use Cases for Azure Stack HCI

Five Steps to Modernizing Your Data

Five Steps to Simplify Your Data Mart and BI Solution

Five Ways to Amplify Power BI with Azure Synapse Analytics

Get hands-on experience with Kubernetes on Azure

Get up and running with Kubernetes

Guide to Datacenter Modernization Through Azure Stack HCI

Hands-On Linux Administration on Azure

How to Choose the Right Azure Services for Your Applications—It’s Not A or B

Invest in App Innovation to Stay Ahead of the Curve

Learn Azure in a Month of Lunches, Second Edition

Limitless Analytics with Azure Synapse

Manage your network more effectively with the Azure Networking Cookbook

Maximize Your Power BI, Tableau, and Qlik Investments

Migrate your SAP estate to the cloud—securely and reliably—with Azure

Migrating Linux to Microsoft Azure

Migrating to Azure: A Resource Guide for Your Database Migration

Modernize Your Applications with Azure Spring Apps

Modernize Your Business-Critical Systems and Applications with the Cloud

Modernize Your Java Apps

Modernize Your Web Apps: Five stories of creating better customer experiences through migration

Professional Azure SQL Managed Database Administration

Start Innovating with Microsoft Azure SQL: Explore common migration and modernization solutions

Succeeding with AI: How to Make AI Work for Your Business

Take your business to the next level with a multicloud strategy

Technical Guide to Building SaaS Apps on Azure

The Business Value of Microsoft Azure for ISVs

The Developer’s 7-Step Guide to Low-Code App Development

The Path to ISV Success: Build, Scale, and Grow Faster with ISV Success

The Road to Azure Cost Governance

The Ultimate Guide to Windows Server 2022

The Ultimate Guide to Windows Server on Azure

Understanding Cloud Migration Strategies

White papers

Find quick reads and thought leadership on key topics such as cloud security, hybrid environments, and the economics of cloud adoption.

Digital Transformation and the Art of the Possible

Empower IT and data professionals to achieve more with all their data

Harvard Business Review Analytic Services: Drive Agility and Innovation with ERP in the Cloud

IDC: The Business Value of Migrating and Modernizing with Azure

Looking to Make Digital Transformation a Reality? Turn to trusted partners for maximum effect

Power BI Professional’s Guide to Azure Synapse Analytics

The Business Value of Microsoft Azure for SQL Server and Windows Server Workloads

The Culture of Data Leaders

Forrester study:  The Projected Total Economic Impact™ of Microsoft Azure VMware Solution

Get started with an Azure free account

Get 12 months of popular free services, a $200 credit to explore Azure services for 30 days, and more than 55 always-free services.

Contact Microsoft Azure Sales

Start a chat session, call us, or have us call you—your choice.

Microsoft Research AI for Science

AI for Science in Conversation: Chris Bishop and Frank Noé discuss setting up a team in Berlin

Watch Research Summit "Fifth Paradigm of Scientific Discovery" plenary on demand

AI for Science to empower the fifth paradigm of scientific discovery

“Over the coming decade, deep learning looks set to have a transformational impact on the natural sciences. The consequences are potentially far-reaching and could dramatically improve our ability to model and predict natural phenomena over widely varying scales of space and time. Our AI4Science team encompasses world experts in machine learning, quantum physics, computational chemistry, molecular biology, fluid dynamics, software engineering, and other disciplines, who are working together to tackle some of the most pressing challenges in this field.“ 未来十年，深度学习注定将会给自然科学带来变革性的影响。其结果具有潜在的深远意义，可能会极大地提高我们在差异巨大的空间和时间尺度上对自然现象进行建模和预测的能力。为此，微软研究院科学智能中心（AI4Science）集结了机器学习、计算物理、计算化学、分子生物学、软件工程和其他学科领域的世界级专家，共同致力于解决该领域中最紧迫的挑战。 Professor Chris Bishop , Technical Fellow, and Director, AI for Science

Work with us

Senior researcher – machine learning  .

Location : Beijing, China

Technical Program Manager 2 – AI for Science  

Our locations.

Amsterdam, Netherlands

Beijing, China

Berlin, Germany

Cambridge, UK

Redmond, USA

Shanghai, China

• Follow on Twitter
• Like on Facebook
• Follow on LinkedIn
• Subscribe on Youtube
• Follow on Instagram
• Subscribe to our RSS feed

Share this page:

• Share on Twitter
• Share on Facebook
• Share on LinkedIn
• Share on Reddit

Tech support scams adapt and persist in 2021, per new Microsoft research

Jul 21, 2021 | Mary Jo Schrade - Assistant General Counsel, Regional Lead, Microsoft Digital Crimes Unit Asia

• Share on Facebook (opens new window)
• Share on LinkedIn (opens new window)
• Share on Twitter (opens new window)

Tech support scams are a global problem, impacting people of all ages. It started with cold calls, with scammers pretending to be Microsoft employees fraudulently notifying people that they were victims of malware infections or other harmful attacks. This evolved into fake “pop-ups” displayed on people’s computers, again trying to convince them that something was wrong with their computers so the scammers could extract payment for “fixing” fake issues. Today, fraudsters have adapted to evolving technology by using more sophisticated tactics or ploys to victimize users online.

Each month, Microsoft receives about 6,500 complaints from people who’ve been victims of tech support scams, which is down from 13,000 reports in an average month in prior years. But it’s not just Microsoft’s brand that the scammers leverage; fraudsters have pretended to be from a number of other reputable tech companies and service providers. To measure the scope of this problem globally, Microsoft commissioned YouGov for a new 2021 survey across 16 countries [1] to look at tech support scams and their impact on consumers. This is a follow-up to similar surveys Microsoft fielded in 2018 and in 2016 .

Results from the 2021 survey reveal that, globally, fewer consumers have been exposed to tech support scams as compared to the 2018 survey. The survey results also show that people are generally more skeptical about tech support calls or pop-up messages, which helps them avoid falling victim to these scams. However, those people who continued with the interaction were more likely to have lost money to the scammers than we saw in our previous survey.

The most interesting results of the survey concern the number of consumers who have experienced tech support scams and the demographics of the consumers who continued with the fraudulent interactions.

A few highlights include:

• Three out of five consumers have encountered a tech support scam in the last 12 months.
• One out of six consumers was tricked into continuing with the scam, often leading to victims losing hundreds of dollars to the fraudsters.
• Millennials (aged 24-37) and Gen Zers (aged 18-23) have the highest exposure to tech support scams.
• One out of 10 millennials and one out of 10 Gen Zers that encountered a scam fell for it and lost money.
• Among those who continued with a scam, the most common issue experienced during the interaction was computer problems (30%), followed by compromised passwords (23%) and fraudulent use of credit/debit/store cards (18%).

Globally, those who lost money reported higher engagement in risky online activities and overestimated their abilities with respect to using computers and the internet. Similar to the 2018 results , we’re seeing younger people fall prey to tech support scams more often, in particular Gen Zers and millennials, as well as males. This also correlated to higher engagement than older generations in riskier online activities, such as using torrent sites and sharing email addresses in exchange for content.

Sensitive financial information continues to be at risk: While scammers most commonly asked consumers to download software or go to a website (with 30% of victims who did so reporting subsequent computer problems), the proportion of consumers asked for their government-issued identification number (e.g. Social Security number) increased since 2018, and 16% were asked to go to their banking website during the session. Not surprisingly, there was an increase in the number of consumers reporting fraudulent use of credit/debit cards, accounting for the increase in money lost. ​

Tech support fraud has evolved from pure cold calling to a more sophisticated infrastructure that leverages affiliate marketers to deliver professional-looking pop-ups to consumers, prompting them to contact fraudulent call centers. We also see scammers using email, search engine optimization (SEO) and social engineering tactics to lure victims. These tactics have served to expand an enterprise model that is easily replicable, with perpetrators sharing resources, including referrals to call centers, leads and payment processors.

Once they engage with potential victims, the scammers steal personal and financial information as well. Those who fell victim paid at least $200 on average, and many victims faced repeat scam interactions. Quite a few unlucky victims even lost thousands of dollars for fake tech support to fix nonexistent computer issues. [2]   Some scammers even installed malware on their “customers’” computers, allowing them to maintain access to the computers even after these victims believed their remote access sessions had ended.

Microsoft’s Digital Crimes Unit (DCU) is working to help combat the problem by partnering with law enforcement, strengthening our technology and educating consumers. Microsoft has been fighting against tech support scams since 2014, when we took our first major strike against online fraudsters with a civil lawsuit in U.S. federal court. Since then, Microsoft has filed its own civil actions and has supported law enforcement officials to take legal action against scammers in the U.S., Asia and Europe.

The DCU works to combat tech support scams by: 1) investigating tech support fraud networks and referring cases to law enforcement as appropriate; 2) strengthening our products and services to better protect consumers from various fraudulent tactics; and 3) educating consumers about this type of fraud by providing guidance and resources on how to identify, avoid and report them.

We hope releasing these findings will help raise awareness and educate consumers about how to protect themselves from tech support scams. Learn more about how to protect yourself from tech support scams here.

[1] Microsoft commissioned YouGov PLC to conduct the survey in 2021. All figures, unless otherwise stated, are from YouGov PLC. Total sample size is 16,254 adults across 16 countries (just over 1,000 per country). Fieldwork was undertaken between April 12, 2021 through May 7, 2021. The survey was executed via YouGov’s online panels and the figures have been weighted and are representative within each country for adults aged 18+. The countries include Australia, Brazil, Canada , Colombia, Finland, France, Germany, India, Japan, Mexico , Netherlands, Singapore, Spain , Switzerland , United Kingdom and the United States – countries in bold are new in 2021. YouGov was not commissioned for the 2016 and 2018 studies, so all year-over-year comparisons are based on provided data.

[2] As provided in customer reports to Microsoft.

Tags: cybercrime , cybersecurity , Digital Crimes Unit , malware , tech support scams

• Check us out on RSS

Chemistry: Technical Reports

• Chemistry Home
• Databases and Preprint Servers
• Dictionaries and Handbooks
• Encyclopedias
• Patents & Trademarks
• Laboratory Safety
• Theses and Dissertations
• Technical Reports
• KiltHub Repository This link opens in a new window

About Technical Reports

Technical reports online.

• NTIS - now the National Technical Reports Library (NTRL) This link opens in a new window Describes government technical reports from the U.S. and other countries. Good for locating reports that one should be able to obtain for free from NASA, DOD, DOE, EPA and other agencies. NTRL has the full text of more than 800,000 technical reports.

Free computer science citation database with some full text available. Lists the most frequently cited authors and documents in computer science, as well as impact ratings. Also provides algorithms, metadata, services, techniques, and software.

Technical Report Collections at Other Universities

University of Maryland - a U.S. Government Document Depository Library for scientific and technical reports from several agencies

Indiana University Computer Science Department

Massachusetts Institute of Technology (MIT)

Stanford University

U.C. Berkeley

University of Washington Computer Science & Engineering Department

Carnegie Mellon Technical Reports

• Computer Science
• Human-Computer Interaction Institute (HCII)
• Information Technology Center (ITC)
• Institute for Complex Engineered Systems (ICES) [formerly: Engineering Design Research Center ]
• Institute for Software Research (ISR)
• Language Technologies Institute (LTI)
• Machine Learning Department [formerly: Center for Automated Learning and Discovery]
• Parallel Data Laboratory (PDL)
• Robotics Institute
• Software Engineering Institute (SEI)
• Statistics Department
• Philosophy Department

U.S. Government Public Technical Reports

A concise list of government agencies with free access to their technical reports:  

Defense Technical Information Center (DTIC)

DTIC helps the Department of Defense (DoD) community access pertinent scientific and technical information to meet mission needs more effectively.

Information Bridge (U.S. Department of Energy)

Provides free public access to over 230,000 full-text documents and bibliographic citations of Department of Energy (DOE) research report literature.  Documents are primarily from 1991-present and were produced by DOE, the DOE contractor community, and/or DOE grantees.

Technical Report Archive and Image Library (TRAIL)

A collaborative project to digitize, archive, and provide persistent and unrestricted access to federal technical reports issued prior to 1975.

Army Corps of Engineers Research and Development Center (CRREL)

The results of CRREL's research projects are published in a technical report series covering topics of interest to Civil and Environmental Engineers.  Reports from 1995 to present are available, as well as some older ones.

NASA Technical Reports Server (NTRS)

1920–present.  Indexes technical reports, conference papers, journal articles, and other publications sponsored by the National Aeronautics and Space Administration (NASA) and its predecessor, the National Advisory Committee for Aeronautics (NACA).  NACA Reports, Technical Notes, and Technical Memoranda are available in fulltext from 1917–1958.  Some NASA reports are fulltext.

Environmental Protection Agency (EPA)

Fulltext of more than 7,000 archival and current EPA documents.

Specialized Technical Reports

Jet Propulsion Laboratory (Caltech)

Lawrence Berkeley National Laboratory (LBNL)

Lawrence Livermore National Laboratory (LLNL)

Los Alamos Technical Reports

IBM Research -Technical Paper Search

Hewlett Packard Labs Technical Reports

Microsoft Research Technical Reports

• << Previous: Theses and Dissertations
• Next: KiltHub Repository >>
• Last Updated: Jan 9, 2024 3:06 PM
• URL: https://guides.library.cmu.edu/Chemistry

What’s next: Microsoft Build continues the evolution and expansion of AI tools for developers

• Share on LinkedIn (opens new window)
• Share on Facebook (opens new window)
• Share on Twitter (opens new window)

Groundbreaking advances in AI are having a profound impact on Microsoft and on the developers using our technology to improve efficiency, enhance customer experiences and make new breakthroughs.

Over the past year, we’ve built Microsoft Copilot and released more than 150 updates to it, and we’ve developed the Copilot stack, which takes everything we’ve learned so far and lets developers build their own copilots.

In addition, over the past two years, GitHub Copilot has become the most widely adopted AI developer tool, with 1.8 million paid subscribers.

And just a day ago, we introduced a new category of Copilot+ PCs, the fastest, most AI-ready PCs ever built.

We are the industry leader in AI, and that’s why, as we begin our flagship event for developers, Microsoft Build, there’s one question on everyone’s mind:  What’s next?

On Monday, we introduced the world to a  new class of Windows PCs , Copilot+ PCs. These devices are designed to allow developers to deliver differentiated AI experiences on the edge, and, together with the  great features that we are announcing at Build , make Windows the most open platform for AI and the best place for developers.

At Build, we also are announcing further branches of this technology revolution, including:

• How Microsoft Fabric is helping developers and customers leverage data in motion, or digital information being transported within or between computer systems, to build intelligent apps.
• New frontier models that allow developers to explore multimodal capabilities supporting text, images, video and other types of data in their AI applications, evolving beyond text prompts and completions.
• Partnerships that show how AI can impact the future of myriad industries, including education.
• And how an open and flexible cloud-based platform, built on a foundation of safe and responsible AI, can help developers innovate.

Microsoft Build  is always an exciting time for us. About 200,000 people have registered to join us for three days of technical learning and community connection, with 4,000 attending in person in Seattle. Attendees can choose from more than 300 sessions, demos and expert and instructor-led labs from Microsoft and our partners. For those not able to attend the live event, most content will be available on demand. In total, we are announcing about 60 new products and solutions at the event.

With that introduction to Build in mind, let’s explore some of the news and announcements.

In-the-moment decision making with Real-Time Intelligence

For the most efficient AI apps, businesses need to be able to qualify, analyze and organize data at ingestion. This has proven to be a difficult step. The new  Real-Time Intelligence  within Microsoft Fabric provides an end-to-end Software as a Service (SaaS) solution that empowers customers to act on high-volume, time-sensitive and highly granular data to make faster and more informed business decisions.

Real-Time Intelligence ,  now in preview , can be helpful for analysts with simple low/no-code experiences, and it also can benefit pro developers with code-rich user interfaces. For example, racing team Dener Motorsport has been using Microsoft Fabric to support real-time analytics, storage and reporting, enabling them to sustain optimal performance and maintain the cars in good repair, which can keep drivers safer. Dener plans to use Real-Time Intelligence as part of their winning strategy to gain insights at ingestion during races.

Building apps requires a level of flexibility, customization and efficiency to make it feasible for developers. The new Microsoft Fabric  Workload Development Kit  makes this possible by enabling independent software vendors (ISVs) and developers to extend applications within Fabric, creating a unified user experience.

GitHub Copilot experience leans into extensibility

GitHub is introducing the first set of  GitHub Copilot extensions , developed by Microsoft and third-party partners, in private preview. These additions allow developers and organizations to  customize their GitHub Copilot experience  with their preferred services like Azure, Docker, Sentry and more directly within GitHub Copilot Chat.

GitHub Copilot for Azure, one of the extensions from Microsoft, showcases how building in natural language with a broader range of capabilities can propel development velocity. Using the extension through Copilot Chat, developers can explore and manage Azure resources, while also troubleshooting issues and locating relevant logs and code.

New frontier models and multimodal capabilities in Azure AI

GPT-4o, OpenAI’s newest flagship model, is now available in Azure AI Studio and as an API. This groundbreaking multimodal model integrates text, image and audio processing to set a new standard for generative and conversational AI experiences.

We also announced that Phi-3-vision , a  new multimodal model  in the Phi-3 family of AI small language models (SLMs) developed by Microsoft, is now available in Azure. Phi-3 models are powerful, cost-effective and optimized for personal devices. Phi-3-vision offers the ability to input images and text and receive text responses. For example, users can ask questions about a chart or ask an open-ended question about specific images.

Developers can experiment with these state-of-the-art frontier models in the Azure AI Playground, and they can start building with and customizing with the models in Azure AI Studio.

Microsoft, Khan Academy using AI to empower educators

Microsoft and Khan Academy are announcing a  multi-faceted partnership  to turn the transformative potential of AI into reality.

To start, Microsoft is enabling Khan Academy to offer all K-12 U.S. educators free access to Khanmigo for Teachers, an AI-powered teaching assistant that frees up teachers’ time so they can focus on what matters most – engaging with and supporting their students. Microsoft is donating access to Azure AI-optimized infrastructure to increase the availability of Khanmigo for Teachers, which will now be powered by Azure OpenAI Service.

Khan Academy is collaborating with Microsoft to explore opportunities to improve math tutoring in an affordable, scalable and adaptable manner with a new version of Phi-3, a family of SLMs developed by Microsoft. They also plan to bring more Khan Academy content into Copilot and Teams for Education, expanding resources for learners.

New partnership with Cognition AI

Microsoft and  Cognition   will bring Cognition’s autonomous AI software agent, Devin, to customers to help them with complex tasks such as code migration and modernization projects. As part of the agreement, Devin will be powered by Azure. Cognition AI is an applied AI lab building end-to-end software agents to help developers achieve more.

Powerful new virtual machines help fuel AI adoption

Microsoft has a unique systems approach to AI infrastructure, which includes hardware and software from Microsoft and our partners, all optimized to run AI workloads at scale and fine-tuned for customer needs. We are the first cloud provider to bring AMD’s leading MI300X AI accelerator chip to power customers’ AI training and inferencing needs, with the general availability of the  Azure ND MI300X v5  virtual machine series  optimized for demanding AI and high-performance computing (HPC) workloads like Azure OpenAI Service.

Following the launch of Azure Cobalt 100, Microsoft’s first custom-designed compute processor, the company is announcing a preview of new  Cobalt 100 Arm-based virtual machines  (VMs), based on the company’s custom silicon series announced in November 2023.  Cobalt 100 Arm-based VMs  are the first generation of VMs to feature Microsoft’s new Cobalt processor, custom-built on an Arm architecture, and optimized for efficiency and performance when running general-purpose and cloud-native workloads. Customers can expect up to 40% improved performance compared to comparable Azure VMs.

The evolution of Copilot

Copilot has been a game-changer for many people since it was first released. Using modern AI and large language models (LLMs) like Open AI’s GPT-4, copilots across Microsoft products have assisted people with complex tasks, serving as a personal, behind-the-scenes AI assistant.

Now, we are introducing  Team Copilot , the expansion of Copilot for Microsoft 365 from a behind-the-scenes, personal AI assistant to a  new, valuable member of your team . You will be able to invoke Copilot where you collaborate – in Teams, Loop, Planner and more. Team Copilot can be a meeting facilitator in meetings, managing the agenda, tracking time and taking notes. It can act as a collaborator in chats by surfacing important information, tracking action items and addressing unresolved issues. It can serve as a project manager to help ensure every project runs smoothly and notify the team when their input is needed. These initial experiences, coming in preview later this year, will enable us to learn, iterate and refine as we enter a new phase of innovation where Copilot begins to take more action on behalf of individuals and teams.

Microsoft   Copilot Studio  is introducing new agent capabilities, empowering developers to build copilots that can proactively respond to data and events, tailored to specific tasks and functions. Copilots built with this new category of capabilities can now independently manage complex, long-running business processes by leveraging memory and knowledge for context, reason over actions and inputs, learn based on user feedback and ask for help when they encounter situations that they don’t know how to handle. Users can now put Copilot to work for them – from IT device procurement to customer concierge for sales and service.

Copilot extensions , including plugins and connectors, allow customers to enhance Microsoft Copilot by connecting it to new data sources and applications, expanding its functionality.

We’re excited about these announcements and all the other updates and features being unveiled at Build. For more information,  today you can watch keynotes  from Microsoft Chairman and CEO Satya Nadella, Executive Vice President of Experiences and Devices Rajesh Jha and Chief Technology Officer Kevin Scott.

On Wednesday,  you can watch keynotes  from Executive Vice President of Cloud and AI Scott Guthrie and friends. Additionally, you can explore all the news and announcements in the  Book of News .

Related Posts

Introducing Copilot+ PCs

Microsoft and LinkedIn released 2024 Work Trend Index: three out of four people use AI at work

The focus has shifted to cyber security

Embracing AI Transformation: Insights from Kalin Dimtchev, Country Manager for Microsoft Adriatics

Microsoft Copilot for Security generally available on April 1, with new capabilities

• Check us out on RSS

Share this page:

Help | Advanced Search

Computer Science > Computer Vision and Pattern Recognition

Title: sora: a review on background, technology, limitations, and opportunities of large vision models.

Abstract: Sora is a text-to-video generative AI model, released by OpenAI in February 2024. The model is trained to generate videos of realistic or imaginative scenes from text instructions and show potential in simulating the physical world. Based on public technical reports and reverse engineering, this paper presents a comprehensive review of the model's background, related technologies, applications, remaining challenges, and future directions of text-to-video AI models. We first trace Sora's development and investigate the underlying technologies used to build this "world simulator". Then, we describe in detail the applications and potential impact of Sora in multiple industries ranging from film-making and education to marketing. We discuss the main challenges and limitations that need to be addressed to widely deploy Sora, such as ensuring safe and unbiased video generation. Lastly, we discuss the future development of Sora and video generation models in general, and how advancements in the field could enable new ways of human-AI interaction, boosting productivity and creativity of video generation.

Submission history

Access paper:.

• HTML (experimental)
• Other Formats

References & Citations

• Google Scholar
• Semantic Scholar

1 blog link

Bibtex formatted citation.

Bibliographic and Citation Tools

Code, data and media associated with this article, recommenders and search tools.

• Institution

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs .

Featured Article

Microsoft dodges UK antitrust scrutiny over its Mistral AI stake

But its inflection hires and amazon’s anthropic deal aren’t so cut-and-dry.

Microsoft won’t be facing antitrust scrutiny in the U.K. over its recent investment in French AI startup, Mistral AI , with the country’s Competition and Markets Authority (CMA) on Friday concluding that the partnership “does not qualify for investigation under the merger provisions of the Enterprise Act 2002.”

The decision comes three weeks after the CMA revealed a trio of early-stage probes into Amazon and Microsoft’s various AI investments and partnerships, including the Redmond-based company’s $16 million investment in Mistral AI , an OpenAI rival working on large language models. Shortly after, Microsoft hired the  team behind Inflection AI , another OpenAI rival, essentially gutting the startup.

Elsewhere, the CMA said it was also poking at Amazon’s $4 billion investment in Anthropic, a U.S.-based AI company working on large language models.

Big Tech and the quasi-merger

There has been growing scrutiny of Big Tech’s latest tactic to dodge regulatory oversight by pursuing “ quasi-mergers ,” through which they seek to secure control over new technologies without buying startups outright. This might be through making investments, procuring seats on boards, hiring founding teams and so on.

Early in 2024, the Federal Trade Commission (FTC)  launched investigations  into Alphabet, Amazon and Microsoft’s investments in emerging AI firms to establish whether the “partnerships pursued by dominant companies risk distorting innovation and undermining fair competition.”

The CMA’s efforts are part of that same regulatory push. Two of its recently announced “invitations to comment” are still ongoing, and may lead to formal in-depth probes. Still, it’s telling that the CMA is throwing out the Mistral AI case on the grounds that it doesn’t “qualify” for investigation under existing rules.

Alex Haffner, competition partner at U.K. law firm Fladgate, says this finding suggests that the structure of Microsoft’s partnership with Mistral AI doesn’t grant the bigger company sufficient rights or influence, at least as it relates to M&A regulation. Ultimately, it was a minority investment into a double-unicorn that had closed a $415 million round just a few months earlier .

“In so doing, the decision vindicates Microsoft’s stated position on the tie-up,” Haffner said.

This “stated position” was that making a small investment isn’t enough to procure meaningful clout in the future direction of an up-and-coming AI startup. Microsoft would effectively own less than 1% of Mistral AI when its investment converts to equity at the French startup’s next funding round.

A Microsoft spokesperson said at the time of the CMA’s initial probe announcement:

“We remain confident that common business practices such as the hiring of talent or making a fractional investment in an AI startup promote competition and are not the same as a merger.” Microsoft spokesperson, April 2024

While the CMA maintains that Big Tech could be adopting new methods to protect themselves from antitrust scrutiny, it has now confirmed that Microsoft hadn’t acquired any “material influence on Mistral AI’s commercial policy.”

“The CMA has considered information submitted by Microsoft and Mistral AI, together with feedback received in response to its invitation to comment,” a CMA spokesperson said. “Based on the evidence, the CMA does not believe that Microsoft has acquired material influence over Mistral AI as a result of the partnership and therefore does not qualify for investigation.”

Pollination works

Just last month, the CMA sounded an alarm over Big Tech’s waxing influence on the advanced AI market, expressing concerns over the growing connection and concentration between developers in the snowballing generative AI space. But the CMA has now said that at least one of the deals on its radar doesn’t qualify for investigation, suggesting that Big Tech’s tactics to pollinate the AI ecosystem far and wide might be working to a degree.

But that still leaves two more outstanding cases: Amazon’s gargantuan investment in Anthropic, and Microsoft’s hiring of key Inflection personnel. Could we expect a similar outcome there?

“The CMA has concluded that the arrangements between Microsoft and Mistral are not sufficient to give Microsoft ‘material influence’ over Mistral, which is the relevant jurisdictional test,” Haffner said. “Time will tell, but the assumption is therefore that the application of the test is more clear-cut here than with the other AI partnerships under investigation by the CMA.”

It’s certainly not as cut-and-dry. Anthropic got Amazon’s biggest venture investment to date, constituting more than half of the $7.6 billion the AI company has raised since its inception three years ago. And while Inflection technically still exists, Microsoft scooped up its founders and various key colleagues — in many ways, that was as good as an acquisition.

And let’s not forget about the CMA’s other separate, but related, ongoing case looking at Microsoft’s close ties with OpenAI. The regulator launched a formal “invitation to comment” aimed at relevant stakeholders in the AI and business spheres last year, and the European Commission (EC)  followed suit in January .

So we probably shouldn’t make too many conclusions about the other pending cases based on today’s news.

“That the CMA has only confirmed the conclusions of the Mistral investigation is interesting, as it leaves open the position on the other two deals, as well as the CMA’s ongoing investigation into Microsoft’s role in the OpenAI project,” Haffner said. “Overall, therefore, it is clear that the competition authorities are continuing to engage very closely with developments in the AI sector, and we can expect several more announcements by the CMA in the near future as to the outcome of their ongoing workstreams in this space.”

We’re launching an AI newsletter! Sign up  here  to start receiving it in your inboxes on June 5.

More TechCrunch

Get the industry’s biggest tech news, techcrunch daily news.

Every weekday and Sunday, you can get the best of TechCrunch’s coverage.

Startups Weekly

Startups are the core of TechCrunch, so get our best coverage delivered weekly.

TechCrunch Fintech

The latest Fintech news and analysis, delivered every Sunday.

TechCrunch Mobility

TechCrunch Mobility is your destination for transportation news and insight.

Elon Musk’s xAI raises $6B from Valor, a16z, and Sequoia

Elon Musk’s AI startup, xAI, has raised $6 billion in a new funding round, it said today, in one of the largest deals in the red-hot nascent space, as he…

Indian EV startup Zypp Electric secures backing to fund expansion to Southeast Asia

Indian startup Zypp Electric plans to use fresh investment from Japanese oil and energy conglomerate ENEOS to take its EV rental service into Southeast Asia early next year, TechCrunch has…

A venture capital firm looks back on changing norms, from board seats to backing rival startups

Last month, one of the Bay Area’s better-known early-stage venture capital firms, Uncork Capital, marked its 20th anniversary with a party in a renovated church in San Francisco’s SoMa neighborhood,…

Families of Uvalde shooting victims sue Activision and Meta

The families of victims of the shooting at Robb Elementary School in Uvalde, Texas are suing Activision and Meta, as well as gun manufacturer Daniel Defense. The families bringing the…

Y Combinator’s Garry Tan supports some AI regulation but warns against AI monopolies

Like most Silicon Valley VCs, what Garry Tan sees is opportunities for new, huge, lucrative businesses.

How Maven’s AI-run ‘serendipity network’ can make social media interesting again

Everything in society can feel geared toward optimization – whether that’s standardized testing or artificial intelligence algorithms. We’re taught to know what outcome you want to achieve, and find the…

Women in AI: Miriam Vogel stresses the need for responsible AI

Miriam Vogel, profiled as part of TechCrunch’s Women in AI series, is the CEO of the nonprofit responsible AI advocacy organization EqualAI.

What are Google’s AI Overviews good for?

Google has been taking heat for some of the inaccurate, funny, and downright weird answers that it’s been providing via AI Overviews in search. AI Overviews are the AI-generated search…

The ups and downs of investing in Europe, with VCs Saul Klein and Raluca Ragab

When it comes to the world of venture-backed startups, some issues are universal, and some are very dependent on where the startups and its backers are located. It’s something we…

Scarlett Johansson brought receipts to the OpenAI controversy

Welcome back to TechCrunch’s Week in Review — TechCrunch’s newsletter recapping the week’s biggest news. Want it in your inbox every Saturday? Sign up here. OpenAI announced this week that…

Deal Dive: Can blockchain make weather forecasts better? WeatherXM thinks so

Accurate weather forecasts are critical to industries like agriculture, and they’re also important to help prevent and mitigate harm from inclement weather events or natural disasters. But getting forecasts right…

Spyware app pcTattletale was hacked and its website defaced

pcTattletale’s website was briefly defaced and contained links containing files from the spyware maker’s servers, before going offline.

Synapse, backed by a16z, has collapsed, and 10 million consumers could be hurt

Synapse’s bankruptcy shows just how treacherous things are for the often-interdependent fintech world when one key player hits trouble. 

Women in AI: Sarah Myers West says we should ask, ‘Why build AI at all?’

Sarah Myers West, profiled as part of TechCrunch’s Women in AI series, is managing director at the AI Now institute.

This Week in AI: OpenAI and publishers are partners of convenience

Keeping up with an industry as fast-moving as AI is a tall order. So until an AI can do it for you, here’s a handy roundup of recent stories in the world…

AI tutors are quietly changing how kids in the US study, and the leading apps are from China

Evan, a high school sophomore from Houston, was stuck on a calculus problem. He pulled up Answer AI on his iPhone, snapped a photo of the problem from his Advanced…

Startups Weekly: Drama at Techstars. Drama in AI. Drama everywhere.

Welcome to Startups Weekly — Haje‘s weekly recap of everything you can’t miss from the world of startups. Sign up here to get it in your inbox every Friday. Well,…

From Plaid to Figma, here are the startups that are likely — or definitely — not having IPOs this year

Last year’s investor dreams of a strong 2024 IPO pipeline have faded, if not fully disappeared, as we approach the halfway point of the year. 2024 delivered four venture-backed tech…

Feds add nine more incidents to Waymo robotaxi investigation

Federal safety regulators have discovered nine more incidents that raise questions about the safety of Waymo’s self-driving vehicles operating in Phoenix and San Francisco.  The National Highway Traffic Safety Administration…

Pitch Deck Teardown: Terra One’s $7.5M Seed deck

Terra One’s pitch deck has a few wins, but also a few misses. Here’s how to fix that.

Women in AI: Chinasa T. Okolo researches AI’s impact on the Global South

Chinasa T. Okolo researches AI policy and governance in the Global South.

Disrupt 2024 early-bird tickets fly away next Friday

TechCrunch Disrupt takes place on October 28–30 in San Francisco. While the event is a few months away, the deadline to secure your early-bird tickets and save up to $800…

Big tech companies are plowing money into AI startups, which could help them dodge antitrust concerns

Another week, and another round of crazy cash injections and valuations emerged from the AI realm. DeepL, an AI language translation startup, raised $300 million on a $2 billion valuation;…

Harlem Capital is raising a $150 million fund

If raised, this new fund, the firm’s third, would be its largest to date.

US pharma giant Cencora says Americans’ health information stolen in data breach

About half a million patients have been notified so far, but the number of affected individuals is likely far higher.

Last day to vote for TC Disrupt 2024 Audience Choice program

Attention, tech enthusiasts and startup supporters! The final countdown is here: Today is the last day to cast your vote for the TechCrunch Disrupt 2024 Audience Choice program. Voting closes…

Signal’s Meredith Whittaker on the Telegram security clash and the ‘edge lords’ at OpenAI 

Among other things, Whittaker is concerned about the concentration of power in the five main social media platforms.

Lucid Motors slashes 400 jobs ahead of crucial SUV launch

Lucid Motors is laying off about 400 employees, or roughly 6% of its workforce, as part of a restructuring ahead of the launch of its first electric SUV later this…

Google invests $350 million in Indian e-commerce giant Flipkart

Google is investing nearly $350 million in Flipkart, becoming the latest high-profile name to back the Walmart-owned Indian e-commerce startup. The Android-maker will also provide Flipkart with cloud offerings as…

Jio Financial unit to buy $4.32B of telecom gear from Reliance Retail

A Jio Financial unit plans to purchase customer premises equipment and telecom gear worth $4.32 billion from Reliance Retail.