OpenAI to Launch Media Manager Tool, Allowing Creators to Control AI Training Data
OpenAI has announced the development of a new tool called “Media Manager,” which will empower creators to specify how their works can be used for AI model training. Scheduled for release in 2025, Media Manager will allow creators to choose which of their works, if any, can be scraped and used as training data for OpenAI’s AI models. This tool addresses concerns raised by artists and creators regarding the use of their work without explicit consent or compensation. Media Manager will provide a more granular level of control than existing methods, such as robots.txt files, allowing creators to exempt specific works from AI data scraping. While Media Manager may be viewed as a positive step towards addressing creators’ concerns, some may remain skeptical due to the extensive use of scraped data in existing AI models. OpenAI maintains that its models are designed for generating new content, not replicating existing works, and employs techniques to prevent repetition. The effectiveness and trustworthiness of Media Manager remain to be seen, but it represents a significant effort by OpenAI to address ethical concerns surrounding AI data scraping and provide creators with greater control over their work.
OpenAI Joins C2PA to Combat AI Disinformation and Promote Content Authenticity
OpenAI has announced its participation in the Coalition for Content Provenance and Authenticity or C2PA, a group dedicated to developing standards for verifying the origin and authenticity of digital content. This move reflects OpenAI’s commitment to addressing concerns about AI-generated disinformation and promoting transparency in the use of AI models. C2PA, founded by Microsoft and Adobe, has established open-source technical standards for labeling AI-generated content, allowing viewers to trace its source and distinguish it from real-world media. OpenAI has already implemented C2PA standards in its DALL-E 3 image generation model and plans to do the same for its upcoming Sora video generation model. In addition to joining C2PA, OpenAI is launching the DALL-E Detection Classifier Researcher Access Program, providing researchers with a tool to identify images generated by DALL-E 3. The company is also partnering with Microsoft to establish a $2 million societal resilience fund to support AI education and understanding among vulnerable populations.
AI Startup Tenyx Achieves Breakthrough with Fine-Tuned Open-Source Language Model
AI startup Tenyx has achieved a significant milestone by fine-tuning Meta’s open-source Llama-3 language model to outperform OpenAI’s GPT-4 in certain domains. This marks the first time an open-source model has surpassed the performance of a leading proprietary model, potentially reshaping the AI landscape. Tenyx’s innovative approach to fine-tuning addresses the issue of “catastrophic forgetting,” where models lose previously learned knowledge when exposed to new data. By selectively updating only a small portion of the model’s parameters, Tenyx enables efficient training on new information without compromising existing capabilities. The company’s commitment to open-source AI is evident in its decision to release the fine-tuned model, Tenyx-70B, under the same license as the original Llama-3. This move promotes accessibility and collaboration within the AI community, potentially leading to further advancements in natural language processing. Tenyx’s breakthrough has significant implications for the AI industry, offering businesses and researchers access to state-of-the-art language models without the limitations of proprietary offerings.
Oracle Unveils Code Assist: AI-Powered Coding Assistant for Enterprise Applications
Oracle has announced Code Assist, an upcoming AI-powered coding assistant designed to accelerate the development of Java, SQL, and Oracle Cloud-based applications. Code Assist will join the growing market of AI coding assistants, aiming to streamline the app development lifecycle and improve developer productivity. The tool will offer various capabilities, including code suggestions, API documentation generation, code summarization, pull request management, and automated testing. Optimized for Java and SQL, Code Assist is specifically tailored for enterprise application development and leverages Oracle’s expertise in these languages. Code Assist will be available as a plugin for popular development environments like JetBrains IntelliJ IDEA and Microsoft Visual Studio Code. Oracle emphasizes the tool’s focus on code consistency, security, and compliance, making it particularly appealing for developers in specialized and regulated industries.
