Time Booster Marketing’s Post

“ Microsoft , opens new tab and OpenAI are probing if data output from the ChatGPT maker's technology was obtained in an unauthorized manner by a group linked to Chinese artificial intelligence (AI) startup DeepSeek, Bloomberg News reported on Tuesday. Microsoft's security researchers observed that, in the fall, individuals they believed to be connected to DeepSeek exfiltrating a large amount of data using the OpenAI's application programming interface (API), the report said. “ From Reuters

“OpenAI, the company behind ChatGPT, says it has proof that the Chinese start-up DeepSeek used its technology to create a competing artificial intelligence model — fueling concerns about intellectual property theft in the fast-growing industry…. …….Security researchers at Microsoft, which has poured billions into OpenAI, discovered last fall that individuals with possible links to DeepSeek were harvesting vast troves of data through OpenAI’s application programming interface, or API, sources told Bloomberg.” Congressional China Hawks will come down hard on export controls and distillation prevention. "As we go forward... it is critically important that we are working closely with the US government to best protect the most capable models," OpenAI statement

OpenAI said Wednesday that it is probing whether Chinese artificial intelligence startup DeepSeek improperly used its data to launch the low-cost AI model that sparked a market selloff Monday. #openai #DeepSeek #ai #Microsoft #chatgpt #aiapps #deepseekai #businessdor https://lnkd.in/dNNArEcT

By Eleanor Olcott via Financial Times OpenAI has found evidence that Chinese AI start-up DeepSeek used its proprietary models to train an open-source competitor, potentially violating intellectual property rights through a process called "distillation." Microsoft and OpenAI blocked accounts suspected to belong to DeepSeek after investigations last year, while industry insiders highlight that distillation is a common practice, making enforcement challenging. DeepSeek's rapid success with its R1 reasoning model—trained at a fraction of the cost of comparable US models—has rattled Silicon Valley, contributing to a 17% drop in Nvidia’s stock amid fears that expensive AI hardware investments may be unnecessary. 𝕏: I'm so sorry I can't stop laughing. OpenAI, the company built on stealing literally the entire internet, is crying because DeepSeek may have trained on the outputs from ChatGPT. They're crying their eyes out. What a bunch of hypocritical little babies. - Ed Zitron (@edzitron) https://lnkd.in/gk3Da5f4

⚡ OpenAI has introduced “ChatGPT Gov”, a revolutionary tool for the public sector. This innovative AI tool aims to transform the way government departments deliver services. For details, read the article…

💥 𝗔𝗜 𝗠𝗢𝗗𝗘𝗟 𝗗𝗜𝗦𝗧𝗜𝗟𝗟𝗔𝗧𝗜𝗢𝗡: 𝗜𝗡𝗦𝗣𝗜𝗥𝗔𝗧𝗜𝗢𝗡 𝗢𝗥 𝗜𝗠𝗜𝗧𝗔𝗧𝗜𝗢𝗡? ©️ In the latest development in the DeepSeek Saga, Microsoft and OpenAI are reportedly investigating whether DeepSeek, exfiltrated #data without its consent, via a process known as Knowledge Distillation or KD, to train its R1 model. 𝗪𝗵𝗮𝘁 𝗶𝘀 𝗞𝗻𝗼𝘄𝗹𝗲𝗱𝗴𝗲 𝗗𝗶𝘀𝘁𝗶𝗹𝗹𝗮𝘁𝗶𝗼𝗻? A machine learning (ML) technique whereby a larger pre-trained model (the Parent Model) transfers learnings to an as yet untrained model (the Student Model). The Student Model asks the Parent Model millions of questions and then, via ML, learns to mimic the reasoning. In the case of DeepSeek-R1, the student has seemingly become the master, with US export controls acting as the mother of invention. 𝗪𝗵𝗮𝘁'𝘀 𝘁𝗵𝗲 𝗮𝗹𝗹𝗲𝗴𝗮𝘁𝗶𝗼𝗻? Microsoft claims that there is substantial evidence that DeepSeek distilled the knowledge of open AI's models using OpenAI's application programming interface (#API), to train its R1 model. Microsoft, OpenAI's largest investor, notified DeepSeek yesterday of its identification of suspicious activity, according to a Bloomberg report. It seems that DeepSeek is yet to comment. 𝗪𝗵𝗮𝘁 𝗵𝗮𝗽𝗽𝗲𝗻𝘀 𝗻𝗼𝘄? If credible, evidence of KD by DeepSeek would infringe OpenAI's Terms of Service, and potentially, its intellectual property (IP) rights. There are of course several barriers to OpenAI taking any legal action, including the obvious jurisdictional challenges. And of course, OpenAI is no stranger to such claims of IP infringement. President Trump's new #AI and #Crypto Czar, David Sacks, in an interview with Fox News yesterday, predicted that in the coming days and weeks, AI model providers will take active steps to prevent unauthorised distillation by third parties. More to follow on this from my team and me at Charles Russell Speechlys- stay tuned. #charlesrussellspeechlys #ai #machinelearning #training #datasets #copyright #intellectualproperty #competition

OpenAI launches multitasking AI. OpenAI is introducing an #AItool capable of managing a user’s computer while simultaneously handling various tasks, such as organizing travel and placing online orders. Named Operator, this application is currently accessible to a limited number of ChatGPT Pro subscribers. #OpenAI refers to it as a "research preview" and seeks feedback from early users to enhance its functionality. The Operator functions as an AI agent, a technology developed by companies like Google and Anthropic. What are your thoughts? How do you see AI agents influencing the future of your industry? We'd love to hear your insights in the comments below. #aiagents #openaioperator

DeepSeek: Microsoft & OpenAI Suspect Borrowed Tech 🕵️♂️💻 News reports suggesting that Microsoft and OpenAI are investigating whether groups linked to DeepSeek stole data from OpenAI. Microsoft security researchers observed groups they believe are linked to DeepSeek exfiltrating large amounts of data from OpenAI several months ago. #Microsoft #OpenAI #DeepSeek #AI #ArtificialIntelligence

Wow this is just another reason why it is a must to have MDM and MTD solutions to protect your employees, your customers, and your business’ reputation. Think twice before you let company devices have acess to these types of tools. Reach out to me and Eric W. if you want some guidance!

We all now know how OpenAI took the world by storm and democratized access to large language models (LLM) and demonstrated to everyone how useful #GenerativeAI can be. We also witnessed several "oh shit" moments when proprietary information ended up in the public through the continuous re-finement/training of the public models. OpenAI with ChatGPT pulled a page from social media platform playbook and offered free access to their tools with the intent to train and optimize their models based on the usage patterns and content from the mostly happy users. They learned a lot about the hallucinations and continue to improve their models. While this is great for the "common" end users, corporate users need to think about privacy and compliance as well. There's a strong case for private instances when using LLMs and GenAI internally in organizations. Regulations, compiance and privacy often dictate that information cannot risk getting commingled with public domain. One of the first to offer a convenient way to build these private instances was AWS with their Bedrock framework and various underlying foundation models including Anthropic's Claude to build a truly private environments. Later Google Gemini and to a degree AzureAI have followed suite. Just recently even Accenture announced their platform built with Meta Llama and Nvidia to offer services for private implementations and some companies have acquired private licenses from OpenAI to run in their private data centers. Organizations and decision makers should ask themselves the question "would I be OK if our data shows up as an answer to our competitor's ChatGPT prompt?" - if the answer is yes, by all means use Copilot and other tools that are available for free or very low cost. If the answer is no, then the first thing you need to do is to ensure training and compliance for your workforce and ideally offer them an environment that is private and safe for everyone to use. There is a huge level of usefulness of these GenAI solutions for various functions such as customer service, marketing, HR, IT etc. that it is detrimental not to leverage them. One final point - various LLMs have different strenghts depending on the use case and their cost highly depends on the way they are being used. It's important to consider the approaches and use cases while balancing the costs. For instance, It might be tempting to use the LLM to create abstracts for complex project plans but there might be a large hidden cost through the token consumption that may make an alternative route a smarter one. Work with a partner who can help you navigate through these questions and guide you to the best implementation roadmap!