#OpenAI
59 Stories
OpenAI’s SearchGPT Falls Short of Expectations, Reveal Early Access Users
Arjun Sha
View quick summary
According to a recent report, OpenAI's SearchGPT, a potential alternative to Google Search, is said to be underwhelming, and it's even less impressive than Perplexity. SearchGPT occasionally hallucinates and struggles to answer local and shopping queries.
OpenAI Close to Releasing Strawberry AI on ChatGPT; ‘Orion’ Could Be GPT-5
Arjun Sha
View quick summary
A new report confirms that OpenAI is close to releasing the Strawberry engine to ChatGPT. Strawberry is an AI engine that can significantly improve reasoning capability and improve performance in math and programming areas. It could launch as early as this fall. Besides that, OpenAI is also training its next-gen 'Orion' AI model, said to be the successor of GPT-4/4o.
Project Strawberry is an internal OpenAI research project that is said to bring human-level reasoning capability. It's related to the Q* project, according to Reuters. Not only that, Project Strawberry can also plan and perform a series of actions unlocking agentic workflows. The project is said to be proficient in solving the hardest math problems.
OpenAI Co-founder Schulman Joins Anthropic, Brockman Takes Extended Leave
Arjun Sha
View quick summary
John Schulman, one of the OpenAI co-founders and alignment head, has left the company to join Anthropic, a rival AI lab. Greg Brockman, OpenAI president and one of the key co-founders of OpenAI, has gone on an extended leave. Schulman says the decision to join Anthropic is a personal one as he wants to work on topics he is interested in.
OpenAI Hesitates to Detect AI-Written Assignments Because It Risks Losing Users
Sagnik Das Gupta
View quick summary
A new WSJ report has highlighted that ChatGPT has had AI-detection tools with 99.9% accuracy in place, ready to release to the public, for a year now. However, the report also suggests that the company has been internally divided on whether or not to make it publicly available. The text watermarking method that OpenAI uses can accurately detect text that's straight away generated by ChatGPT without going through paraphrasing or any such external modifiers. Meanwhile, ChatGPT is also working on detecting the origins of a text via its metadata, which is a bit more complex to achieve. There have also developed C2PA Metadata for AI-generated Images that provides a reliable history of such images. All things considered, OpenAI should really stand its ground and release these tools to the public, and finally stay true to its promise of using AI responsibly.
After months of anticipation, OpenAI has finally unveiled its AI-powered ChatGPT search engine, known as SearchGPT. It's designed to answer fast using authentic sources and clear attribution. Users can also find all the sources and engage with the web results. That said, it's available in a limited prototype which means users have to join the waitlist to receive access.
OpenAI Launches GPT-4o mini Model; Faster and Cheaper Than GPT-3.5 Turbo
Arjun Sha
View quick summary
OpenAI is back with a solid lightweight model that doesn't compromise on performance. The new GPT-4o mini model scores 82% on the MMLU benchmark and costs much less than GPT-3.5 Turbo, Gemini Flash, and Claude 3 Haiku. In fact, it outperforms some of the largest models such as Claude 3 Opus and GPT-4. You can start using the mini model on the free and paid version of ChatGPT starting today.
OpenAI has finally released a statement saying that GPT-4o's Voice Mode is being delayed. It will be available by the end of the next month to a select group of paid users in Alpha. And by the end of this year, all ChatGPT Plus users will get the new feature. The company is also working to bring screen-sharing capability to ChatGPT.
OpenAI’s Former Chief Scientist Ilya Sutskever Launches a New AI Company
Arjun Sha
View quick summary
Ilya Sutskever, the brain behind ChatGPT at OpenAI, left the company last month and now, he has announced a new company called Safe Superintelligence Inc. (SSI). It aims to create a safe superintelligence to ensure safety always remains ahead of capabilities.
Last year, OpenAI fired its CEO Sam Altman, only to appoint him back to the position a couple of days later. Now, some more details from ex-OpenAI board members, Helen Toner and Tasha McCauley, have revealed the actual reasons behind Altman's firing and that has understandably raised some big concerns in the community. From "psychological abuse" to secrecy and lies, Altman has been accused of it all by the ex-board members.
In its blog, OpenAI has revealed that the company is training its next big frontier model and it will unlock the next level of AI capabilities, bringing us closer to AGI. The company has also formed a Safety and Security Committee to evaluate upcoming models and look into critical safety and security decisions.
Last week, OpenAI demoed the 'Sky' voice during the ChatGPT 4o launch. The voice seemed straight out of the movie 'Her', where Scarlett Johansson lent her voice to the AI character 'Samantha'. Now, it turns out Sam Altman, the CEO of OpenAI, was in contact with Scarlett Johansson to voice the new ChatGPT 4o system. However, Johansson had refused the offer, way back in September 2023. Now, Johansson is seeking clarity and transparency as to how the 'Sky' voice was developed. Meanwhile, OpenAI has paused the 'Sky' voice on ChatGPT.
Ilya Sutskever bids farewell to OpenAI and steps down from the role of chief scientist at the company. Ilya says he will be working on a deeply meaningful personal project. Many reports suggest that Ilya was responsible for OpenAI CEO Sam Altman's firing in November 2023.
OpenAI has finally unveiled its next flagship model called GPT-4o which is a truly multimodal model. And it will be available to all ChatGPT users including free and paid users. On top of that, all ChatGPT Plus features are being made available to free users as well. OpenAI says the update will roll out in the next few weeks. Besides that, OpenAI has also launched a desktop app for macOS.
Quelling all rumors, OpenAI has finally announced an event for May 13 at 10 AM PT (10:30 PM IST). It was anticipated that OpenAI would release its ChatGPT search engine challenging Google's dominance. However, Sam Altman says that the company will demo some new stuff that works like magic.
A new report now confirms that OpenAI is going to release its ChatGPT search engine on Monday (May 13). The report also says that the OpenAI search engine will include citations and source real-time information from the web. The search engine has now been indexed by Bing and DuckDuckGo as well.
OpenAI is reportedly working on a new search engine, an alternative to Google Search. And the company may launch it in May, just before Google I/O. Currently, the subdomain doesn't throw a 404 error, but displays "Not found".
YouTube CEO Says OpenAI Could’ve Violated Platform Rules For Sora Training
Arjun Sha
View quick summary
Speaking to Bloomberg, the YouTube CEO, Neal Mohan said that OpenAI would violate YouTube's Terms of Service if the company used YouTube videos to train its Sora model. Google also uses some corpus of YouTube videos to train its Gemini model, but signs licensing contracts with individual creators. Earlier, the CTO of OpenAI, Mira Murati, refused to clarify whether Sora was trained on YouTube videos.
While ChatGPT offers the option to disable model training on your content inside ChatGPT's settings page, it also disables Chat History. This discourages users from disabling model training. That said, OpenAI has a separate privacy portal page where you can request the company to not train its model on your data while keeping the Chat History functionality intact. OpenAI doesn't promote this privacy portal page inside ChatGPT's interface. In fact, a passing mention is found on its FAQ page.
Just like Microsoft Copilot, you can now use ChatGPT without an account. You no longer need to sign in or create an account to use the free version of ChatGPT. The GPT-3.5 model is readily available to all free users including those without an account. Keep in mind, you won't get some features like custom instructions, chat history, chat sharing, and voice conversations without an account.
After introducing Sora, a remarkable text-to-video AI model, OpenAI has previewed its Voice Engine model that can clone voices with a single 15-second audio sample. It can translate voices into different languages with high accuracy. Since there are risks associated with the technology, the company is not releasing the speech synthesis model now. The company encourages society to understand the capabilities of AI and adapt to a new reality.
The ChatGPT maker, OpenAI, is reportedly planning to release the GPT-5 model during the summer of this year. A new report says that OpenAI has already demoed the GPT-5 model to a few enterprise customers and one of the CEOs called it "materially better" than GPT-4. OpenAI is also working on AI agents that can perform complex actions.
OpenAI might be working to release an intermediate GPT-4.5 Turbo model before launching the next-gen GPT-5 model. It was leaked through the OpenAI Blog page which was indexed by Bing and DuckDuckGo. The upcoming model promises better "speed, accuracy, and scalability" than the GPT-4 Turbo model. However, strangely, the captured description says the model will have a knowledge cutoff of up to June 2024.
OpenAI Unravels Some Secrets to Counter Elon Musk’s Lawsuit; Who’s in the Wrong?
Sagnik Das Gupta
View quick summary
Recently, Elon Musk filed a lawsuit against OpenAI, claiming that the company has swerved off track from its original mission. Instead of sticking to its original mission of being a non-profit and making AI systems for human welfare, OpenAI is allegedly working towards making Microsoft more profitable. Dismissing these claims, OpenAI has now unveiled some sensitive emails between the company and Elon Musk to counter Musk's allegations. The emails paint Musk in a different light, suggesting that he wanted complete control over OpenAI, with the company using Tesla as its major source of funding.
AI Models in India to Require Govt Approval; What are the Implications?
Arjun Sha
View quick summary
The Indian IT Ministry (MeitY) has issued a new advisory for tech companies offering AI models and services in India. The advisory asks tech firms to get approval before deploying "untested" AI models in India. The government later clarified that the advisory applies to large companies like Google, OpenAI, Microsoft, etc., and not startups. The advisory also requests companies to embed permanent metadata in generated data to easily identify the first originator.
In a surprising turn of events, Elon Musk has sued OpenAI and its CEO, Sam Altman in the San Francisco Superior Court on Thursday. Musk claims that OpenAI has become a closed-source company and it's only developing new technologies to maximize profits. Further, Musk says that Microsoft is leveraging its power and influencing OpenAI's operations. Musk is seeking an injunction against OpenAI and Microsoft from taking advantage of the AGI technology and cashing in for profits.
OpenAI has unveiled its remarkable text-to-video model, Sora, that can generate AI videos for up to 1 minute. The generated videos are highly detailed and have a resolution of up to 1080p. Unlike existing solutions, Sora-generated videos don't have distortions in the scene, which makes everyone believe that it's trained on gaming engine simulations like Unreal Engine 5. OpenAI says it's a diffusion model built on the transformer architecture. The Sora model is going through safety checks now and currently, it's not available to regular users.
With the new memory addition to ChatGPT, the AI chatbot can now remember who you are, what you like, and your overall preferences. It will make ChatGPT more personal and helpful to users. The memory feature is currently being rolled out to a small portion of ChatGPT free and Plus users. Over the coming days, it will be enabled by default for all ChatGPT users. Keep in mind, your ChatGPT conversations including memories may be used to train and improve OpenAI's AI models.
Sam Altman’s Most Used App Is NOT ChatGPT; Here’s What He Uses All Day!
Satyam Kumar
View quick summary
Sam Altman sat down with Bill Gates in a recent podcast. It was revealed here that it is not ChatGPT which is the most used app by OpenAI's CEO. In fact, it is not any other OpenAI tool either - Sam Altman has revealed he uses Slack the most on his phone. This reveals how busy the CEO is in managing the OpenAI team and its efforts towards artificial general intelligence.
The App Store of GPTs is finally live. OpenAI has rolled out the GPT Store for ChatGPT Plus users after multiple delays. You can explore GPTs for various categories including Writing, Research and Analysis, Programming, Education, Productivity, Lifestyle, and more. Bear in mind, it's not available to free users, and you have to subscribe to ChatGPT Plus to use the GPT Store. OpenAI says that GPT creators can also monetize their GPTs based on the usage.
The GPT Store was delayed multiple times in the last two months, and finally, OpenAI is ready to launch it next week. You will be able to find custom GPTs from various developers, and users and use them if you want to. The store will be public and users would be able to interact with the custom GPT effortlessly. However, to build GPTs, you need to subscribe to ChatGPT Plus.
The New York Times Sues OpenAI and Microsoft Over Copyright Infringement
Arjun Sha
View quick summary
The New York Times has sued OpenAI and Microsoft for using millions of Times' articles to train its AI models without paying any licensing fee. In the lawsuit, the NYT has alleged that GPT-4 produces verbatim output from the NYT articles with minor or no changes at all without any attribution. The NYT is seeking "billions" in damages from both OpenAI and Microsoft over the unlawful use of its journalistic work. The major American newspaper had tried to negotiate a deal back in April, but it didn't go anywhere.
ChatGPT was going to get a GPT store this month. With this, people could make their own custom GPTs, and then based on how many people use it, there is a potential to earn money from custom GPTs too. However, OpenAI's ChatGPT team has just confirmed that the GPT Store is now delayed till early 2024.
Within a year, ChatGPT has become a full-fledged AI chatbot with knowledge cut-off of up to April 2023. It supports a multitude of features including image analysis, GPTs, voice chat, web browsing, code interpreter, plugins, custom instructions, data analysis, and more. ChatGPT also has apps for both Android and iOS. It's going to get access to the latest GPT-4 Turbo model as well.
We recap the events of Sam Altman's ousting from OpenAI and his return to the company within a week. Chaos ensued when OpenAI revealed on November 18 that Sam Altman has been fired. This was followed by Greg Brockman quitting OpenAI. Then, we learned former Twitch CEO Emmet Shear was hired as the interim OpenAI CEO and Sam Altman was moving to Microsoft to set up a new AI research division. Now, on November 22, Altman has been reinstated as the OpenAI CEO with a new board.
Sam Altman was kicked by the OpenAI board. The co-founder left after that, along with 700 out of 770 employees in OpenAI threatening to do the same (and join Microsoft). Now, he is back as OpenAI's CEO!
After OpenAI's decision to hire a second interim CEO, Microsoft announced it will hire Sam Altman (and other AI employees) for a new research team at its company. Altman will reportedly lead the company's new advanced AI research team.
OpenAI Fires Sam Altman as CEO; Cites Lack of Candor
Upanishad Sharma
You Can Now Create Your Own Custom ChatGPT and It’s Easy Too
Vanshika Malhotra