<div>
  <p>
  </p>
  <ul>
    <li>
      <p>
        Ilya Sutskever, OpenAI's co-founder and chief scientist, has announced his departure from the company claiming personal endeavors as the reason.
      </p>
    </li>
    <li>
      <p>
        Jakub Pachocki, the company&rsquo;s current director of research, will replace Sutskever. Jakub reportedly has led the development of GPT-4 and OpenAI Five and fundamental research in large-scale RL and deep learning optimization.
      </p>
    </li>
  </ul>
  <p>
  </p>
</div>


<div>
  <p>
    <a href="https://techcrunch.com/2024/05/14/ilya-sutskever-openai-co-founder-and-longtime-chief-scientist-departs/">
      TechCrunch
    </a>
  </p>
  <p>
    <a href="https://www.theverge.com/2024/5/14/24156920/openai-chief-scientist-ilya-sutskever-leaves">
      The Verge
    </a>
  </p>
  <p>
  </p>
</div>


Generative AI Applications

<div>

<p>OpenAI's co-founder and chief scientist, Ilya Sutskever, has announced his departure from the company, citing personal reasons. Jakub Pachocki, the current director of research at the company, will take his place. Pachocki has reportedly led the development of GPT-4 and OpenAI Five and has significant contributions to fundamental research in large-scale RL and deep learning optimization.</p>

</div>

OpenAI co-founder and chief scientist Ilya Sutskever announces departure from company

Foundation Models

<div>
 
 <p>OpenAI's co-founder and chief scientist, Ilya Sutskever, is leaving the company for personal endeavors. The current director of research, Jakub Pachocki, is set to replace him. Pachocki has led the development of important projects like GPT-4 and OpenAI Five and has conducted fundamental research in large-scale RL and deep learning optimization.</p>
 
</div>

PROMINENT_CATEGORY

<div>
 
 <p>OpenAI has introduced GPT-4o (GPT omni), an advanced model with capabilities to process text, audio, and image inputs and outputs, aimed to improve human-computer interaction. The firm states that this model equals the performance of GPT-4 Turbo in English text and code processing, with enhanced efficiency for non-English languages. It is also said to be twice quicker and 50% more economical than GPT-4 Turbo. </p>

<p>This model also reportedly responds to audio inputs around 320 milliseconds, which approaches human response time during dialogues.</p>

<p>At present, GPT-4o offers text and image features. Initial audio outputs remain restricted to predefined voices to comply with safety procedures, while advancements are slated for release in the weeks ahead.</p>

<p>Currently, developers can use GPT-4o in the API as a text and image processing model.</p>

</div>

OpenAI launches GPT-4o, a model to facilitate natural human-computer interaction

ANALYST_QUICK_TAKE

<div>
 
 <p>OpenAI has launched a new model named "GPT-4o" that is now available for ChatGPT Plus and Team users and will soon be accessible to Enterprise users. The "o" in GPT-4o stands for "omni".</p>
 <p>Featuring support for 50 languages, this model displays multimodal functionality as it integrates text and images. It provides improved intelligence and real-time responsiveness. GPT-4o also enhances the user experience on ChatGPT by understanding emotional nuances and offering improved vision capabilities to handle image- or screen-related inquiries.&nbsp;</p>
 <p>Free users of this model will benefit from limited access to GPT-4 level intelligence, web-integrated responses, and data analysis tools. In addition, they will be able to chat about photos, upload files for assistance, and navigate the GPT Store.</p>
 <p>OpenAI also revealed the release of a new desktop application for macOS that features integrated ChatGPT functionality.</p>
 
</div> <p><b>Analyst QuickTake:<br>
 </b> GPT-4o is a large multimodal model that accepts image and text inputs and produces text outputs. It reportedly exhibits human-level performance on various professional and academic benchmarks. Additionally, features that were previously paywalled, such as memory capability, file and photo uploading, and web searches for timely questions, are now available to free users, allowing a wider range of users to benefit from GPT-4o's capabilities and experience its improved performance and features.</p>

OpenAI launches new model ‘GPT-4o,' desktop app for iOS, and more free features

<div>
 
 <p>California Governor Gavin Newsom announced partnerships with five companies, including OpenAI and Anthropic, for the use of GenAI  tools in public service delivery. The anticipated enhancements pertain to traffic management, road safety, and tax guidance.</p>
 <p>Over six months, internal trials will be undertaken by state workers testing and evaluating these AI tools. The collaborating companies will offer their services for just USD 1 per proposal. The programs are designed so as not to involve any sensitive data, addressing privacy concerns.</p>
 
</div> <p><b>Analyst QuickTake:<br>
 </b> The state of California appears to have a strong foothold in AI technology, with a significant number of AI companies established within its confines. Governor Newsom signed an executive order last year to explore ways to incorporate generative AI, positioning California at the forefront in AI utilization for public service functions.</p>

California partners with firms to deploy GenAI tools for public service delivery

<div>
 <p>OpenAI has introduced a new framework, Model Spec, that is designed to guide the behaviour of future AI models. The framework aims to ensure AI tools comply with laws, minimize information hazards, respect rights, and protect user privacy among other factors.</p>
 <p>The framework is governed by numerous important rules including, obeying the chain of command, abiding by relevant laws, not disseminating hazardous information, respecting the rights of creators, safeguarding personal privacy, and avoiding inappropriate content. A unique feature of this framework is that it allows both companies and users to modify the sensitivity level in AI outputs, thereby controlling the generation of content, such as NSFW, in a context-appropriate manner.</p>
</div>

OpenAI introduces Model Spec framework to guide AI behavior

<div>
 
 <p>OpenAI has announced a strategic partnership and licensing agreement with Dotdash Meredith (DDM), a leading digital and print publisher. The agreement aims to improve the accuracy and reliability of OpenAI's ChatGPT through the inclusion of verified content from DDM.</p>
 <p>The partnership will integrate diverse content from DDM, including trusted recipes, expert-backed health and financial data, popular style and entertainment content, and product evaluations from over 40 media brands. Some of these brands are PEOPLE, Better Homes &amp; Gardens, FOOD &amp; WINE, Verywell, InStyle, and Investopedia.</p>
 <p>In addition to sourcing content from DDM, OpenAI will also include content and links acknowledged to DDM in relevant responses generated by ChatGPT. OpenAI also plans to work with Dotdash Meredith to gain insights from their historical and present data to further enhance the performance of its machine learning model.</p>
 
</div> <p><b>Analyst QuickTake:<br>
&nbsp; </b> OpenAI has been entering numerous partnerships recently with news content publishers. Last month, OpenAI partnered with <a href="https:<br>
//sp-edge.com/updates/28934"> Financial Times </a> to incorporate news content into ChatGPT, and the month before, it formed strategic partnerships with <a href="https:<br>
//sp-edge.com/updates/27567"> French newspaper Le Monde and Spanish media company Prisa Media </a> to access news content through ChatGPT. OpenAI has also made deals with other news organizations to license content to train AI models, including Axel Springer, Politico, and European publications Bild and Welt and The Associated Press.</p>

OpenAI partners with Dotdash Meredith to enhance ChatGPT with media content

<div>

<p>OpenAI has launched a new tool known as Media Manager intended for creators. This tool will enable creators to manage the way their content is used in the development of OpenAI's GenAI model.</p>
<p>Media Manager gives creators the ability to tag their work and dictate how it will be used in the training and research processes of OpenAI's AI systems. The tool also aids in recognizing copyrighted materials like text, images, audio, and video from various sources and ensures they adhere to the creators' set guidelines.</p>
<p>According to OpenAI, the tool will not only offer copyright protection but also broaden the pool of data sources, inclusive of both public and proprietary data obtained through partnerships.</p>

</div> <p><b>Analyst QuickTake:<br>
 </b> This Media Manager development news comes amidst OpenAI’s recent partnerships with <a href="https:<br>
//sp-edge.com/updates/28934"> The Financial Times </a> , <a href="https:<br>
//sp-edge.com/updates/27567"> French newspaper Le Monde, and Spanish media company Prisa Media </a> to access news content through ChatGPT. OpenAI has also partnered with other news organizations to license content to train AI models, including Axel Springer, Politico, and European publications Bild and Welt and The Associated Press.</p>

OpenAI introduces Media Manager to control content use in AI training

<div>
 
 <p>OpenAI has announced a new tool for identifying images produced by its DALL-E generator. This tool reportedly achieves around a 98% success rate in identifying DALL-E 3 images, but performs less accurately when confronted with modified DALL-E 3 images and those created by different AI models.</p>
 <p>Additionally, the DALL-E image generator is set to have watermarking features incorporated, in line with the requirements set out by the Coalition for Content Provenance and Authenticity (C2PA) standards.</p>
 
</div> <p><b>Analyst QuickTake:<br>
 </b> Detecting AI-generated images has become a challenge amidst deepfakes. Last month, Meta and <a href="https:<br>
//sp-edge.com/updates/28333"> Google </a> claimed they plan to label AI-generated media using the C2PA standard.</p>

OpenAI launches tool to detect images generated from DALL-E

INCUMBENTS_PRESENT

<div>
 
 <p>Microsoft and OpenAI have established a USD 2 million fund with the objective of dealing with potential dangers posed by AI and deepfakes, primarily those threatening democratic processes.</p>
 <p>Initiatives envisaged under this fund primarily concentrate on AI and possible exploitation, including the augmentation of "deep fakes" detection, advocating for ethical AI usage and enhancing the public's AI comprehension. In addition, efforts are underway to limit certain AI features during campaign periods, which forms a component of the actions tech giants are taking to prevent AI abuse.</p>
 
</div> <p><b>Analyst QuickTake:<br>
 </b> This initiative could improve societal resilience by promoting a better understanding and literacy of AI among citizens. This would help prevent deception and bolster genuine information, primarily aimed at voters and communities vulnerable to misinformation. Additionally, it intends to spur collaboration and partnerships with organizations that align with their mission.</p>

All Updates

OpenAI co-founder and chief scientist Ilya Sutskever announces departure from company

OpenAI co-founder and chief scientist Ilya Sutskever announces departure from company

OpenAI launches GPT-4o, a model to facilitate natural human-computer interaction

OpenAI launches new model ‘GPT-4o,' desktop app for iOS, and more free features

California partners with firms to deploy GenAI tools for public service delivery

OpenAI introduces Model Spec framework to guide AI behavior

OpenAI partners with Dotdash Meredith to enhance ChatGPT with media content

OpenAI introduces Media Manager to control content use in AI training

OpenAI launches tool to detect images generated from DALL-E

Microsoft and OpenAI launch USD 2 million fund to address deepfake threats