Former OpenAI Staffers Implore Courts to Block What It's Trying to Do
Nearly ten years ago, tech tycoons Sam Altman and Elon Musk launched OpenAI with a promise to develop AI tech to further the good of humanity.
But as the years passed and the market surrounding artificial intelligence grew, so too did the ambitions of OpenAI's executives (Musk flamed out from the venture in 2019, and now runs his own AI company, called xAI.)
Now, the venture that began as a transparent tech nonprofit is quickly turning into a typical Silicon Valley startup — complete with whistleblowers speaking out against the company's foray into the private market.
Earlier this week, a group of former OpenAI employees, law professors, activists, and Nobel Prize winners sent a letter to the California and Delaware attorneys general pleading with them to stop OpenAI from transforming itself into a private company. Though a number of OpenAI's operations have pivoted to a commercial model in recent years — like ChatGPT, which charges a subscription for higher-performing versions — the company's current plan is to restructure itself as an entirely for-profit venture.
"OpenAI may one day build technology that could get us all killed," said former OpenAI employee Nisan Stiennon, alluding to the company's pursuit of Artifical General Intelligence (AGI), the hypothetical point at which machine intelligence matches or exceeds human ability. "It is to OpenAI's credit that it's controlled by a nonprofit with a duty to humanity. This duty precludes giving up that control."
At the moment, most of the titans in the AI industry operate as commercial ventures, as opposed to nonprofits. With this move, OpenAI would simply be joining the rat race — but insiders say the restructuring would bring the company and its frontrunning tech beyond the humanitarian promise that pushed it to the top in the first place.
As a nonprofit, OpenAI's directors have a legal obligation to follow the company's charter, which currently includes the goal "to ensure that artificial general intelligence benefits all of humanity," according to Todor Markov, a former OpenAI employee who now works for Anthropic, which is overseen by a 5-person long-term benefit trust.
"Directors of the [public benefit corporation] would have no such fiduciary duty," Markov explains of the OpenAI transition. "They would be *allowed* to balance [that duty] against shareholder interests, but not *required* to do so... as long as they haven't broken any laws, you have no recourse."
Altman was previously ousted from OpenAI after pulling some shady stunts in 2023, including hiding the release of ChatGPT from his board of directors, and approving enhancements to ChatGPT 4.0 without running them through the company's jointly-organized Safety Board. That "blip," as it came to be known, only lasted five days before Altman was reinstalled, but it remains a black mark on the tech tycoon's reputation.
The question of whether OpenAI is allowed to proceed as a public benefit corporation is up to the two states' attorneys general. Whether it ultimately matters is another question entirely; AGI technology is still a pipe dream at the moment, while a growing body of research suggests it's impossible, at least building on today's technology — which would make the AGI threat of a for-profit OpenAI vastly overblown.
More on AI: Sam Altman Admits That Saying "Please" and "Thank You" to ChatGPT Is Wasting Millions of Dollars in Computing Power
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Bloomberg
42 minutes ago
- Bloomberg
Apple's Upgrades to CarPlay, iPad and Vision Pro Outshine Liquid Glass
Apple's unveiling of iOS 26 and the Liquid Glass interface was the most discussed part of its developers conference, but the real gems were its enhancements to CarPlay, the Vision Pro and the iPad. Also: Delayed Siri upgrades are now targeted for next spring; Apple's AI chief lies low at WWDC; and the company gets cozier with OpenAI. Last week in Power On: Apple's iOS 26 Liquid Glass interface is a prelude to the 20th anniversary iPhone.
Forbes
an hour ago
- Forbes
Doing The Work With Frontier Models: I'll Talk To AI
Artificial Intelligence processor unit. Powerful Quantum AI component on PCB motherboard with data ... More transfers. Within the industry, where people talk about the specifics of how LLMs work, they often use the term 'frontier models.' But if you're not connected to this business, you probably don't really know what that means. You can intuitively apply the word 'frontier' to know that these are the biggest and best new systems that companies are pushing. Another way to describe frontier models is as 'cutting-edge' AI systems that are broad in purpose, and overall frameworks for improving AI capabilities. When asked, ChatGPT gives us three criteria – massive data sets, compute resources, and sophisticated architectures. Here are some key characteristics of frontier models to help you flush out your vision of how these models work: First, there is multimodality, where frontier models are likely to support non-text inputs and outputs – things like image, video or audio. Otherwise, they can see and hear – not just read and write. Another major characteristic is zero-shot learning, where the system is more capable with less prompting. And then there's that agent-like behavior that has people talking about the era of 'agentic AI.' If you want to play 'name that model' and get specific about what companies are moving this research forward, you could say that GPT 4o from OpenAI represents one such frontier model, with multi-modality and real-time inference. Or you could tout the capabilities of Gemini 1.5, which is also multimodal, with decent context. And you can point to any number of other examples of companies doing this kind of research well…but also: what about digging into the build of these systems? At a recent panel at Imagination in Action, a team of experts analyzed what it takes to work in this part of the AI space and create these frontier models The panel moderator, Peter Grabowski, introduced two related concepts for frontier models – quality versus sufficiency, and multimodality. 'We've seen a lot of work in text models,' he said. 'We've seen a lot of work on image models. We've seen some work in video, or images, but you can easily imagine, this is just the start of what's to come.' Douwe Kiela, CEO of Contextual AI, pointed out that frontier models need a lot of resources, noting that 'AI is a very resource-intensive endeavor.' 'I see the cost versus quality as the frontier, and the models that actually just need to be trained on specific data, but actually the robustness of the model is there,' said Lisa Dolan, managing director of Link Ventures (I am also affiliated with Link.) 'I think there's still a lot of headroom for growth on the performance side of things,' said Vedant Agrawal, VP of Premji Invest. Agrawal also talked about the value of using non-proprietary base models. 'We can take base models that other people have trained, and then make them a lot better,' he said. 'So we're really focused on all the all the components that make up these systems, and how do we (work with) them within their little categories?' The panel also discussed benchmarking as a way to measure these frontier systems. 'Benchmarking is an interesting question, because it is single-handedly the best thing and the worst thing in the world of research,' he said. 'I think it's a good thing because everyone knows the goal posts and what they're trying to work towards, and it's a bad thing because you can easily game the system.' How does that 'gaming the system' work? Agrawal suggested that it can be hard to really use benchmarks in a concrete way. 'For someone who's not deep in the research field, it's very hard to look at a benchmarking table and say, 'Okay, you scored 99.4 versus someone else scored 99.2,'' he said. 'It's very hard to contextualize what that .2% difference really means in the real world.' 'We look at the benchmarks, because we kind of have to report on them, but there's massive benchmark fatigue, so nobody even believes it,' Dolan said. Later, there was some talk about 10x systems, and some approaches to collecting and using data: · Identifying contractual business data · Using synthetic data · Teams of annotators When asked about the future of these systems, the panel return these three concepts: · AI agents · Cross-disciplinary techniques · Non-transformer architectures Watch the video to get the rest of the panel's remarks about frontier builds. What Frontier Interfaces Will Look Like Here's a neat little addition – interested in how we will interact with these frontier models in 10 years' time, I put the question to ChatGPT. Here's some of what I got: 'You won't 'open' an app—they'll exist as ubiquitous background agents, responding to voice, gaze, emotion, or task cues … your AI knows you're in a meeting, it reads your emotional state, hears what's being said, and prepares a summary + next actions—before you ask.' That combines two aspects, the mode, and the feel of what new systems are likely to be like. This goes back to the personal approach where we start seeing these models more as colleagues and conversational partners, and less as something that stares at you from a computer screen. In other words, the days of PC-DOS command line systems are over. Windows changed the computer interface from a single-line monochrome system, to something vibrant with colorful windows, reframing, and a tool-based desktop approach. Frontier models are going to do even more for our sense of interface progression. And that's going to be big. Stay tuned.
Tom's Guide
6 hours ago
- Tom's Guide
ChatGPT Projects just got smarter — here's how to use the new tools
OpenAI's new ChatGPT Projects feature just got a huge upgrade, and it's a game-changer for anyone using ChatGPT to manage complex workstreams. Whether you're planning a major event or project, a busy professional or just trying to keep your thoughts organized, Projects gives you a centralized hub where your chats, files and instructions can all live in one focused workspace.I have found it extremely helpful for keeping Custom GPT instructions in one place in case there is another ChatGPT outage. I also use it to keep all my favorite prompts in one place. Here's everything you need to know to get started, including how to create a Project, what it can (and can't) do, and why it just might become your favorite new productivity tool. ChatGPT Projects are like folders for your chats. Each 'Project' lets you: This makes Projects ideal for larger tasks that require ongoing iteration, deeper context or collaboration. I'm currently using it for all the polished images and edits for my current middle grade novel – it's a breeze having everything in one place. To create a new Project: In the left-hand sidebar, click New Project Get instant access to breaking news, the hottest reviews, great deals and helpful tips. Give it a clear, goal-oriented name (e.g., 'College Applications,' 'Business Ideas' or 'Novel Outline') Upload any relevant files and add custom instructions to guide ChatGPT's behavior. This is a great time to use the '3-word-rule' (e.g., 'Act like a UX expert giving me design advice') Instructions you add here will only apply inside this Project — not to your general ChatGPT usage elsewhere. You can drag and drop existing chats into a project or use the menu next to any chat to select Move to project or Create new project. Once a chat is inside a project, it will take on that project's custom instructions and can reference any files you've uploaded. This creates a seamless thread of context that helps ChatGPT deliver smarter, more consistent responses. To remove a chat from a project, just drag it out or choose Remove from the chat's something worth considering: although there is a separate Image Library, I like to use Projects to keep the images together for a specific project. That way everything stays organized and in one place. Projects support a wide range of tools, making them a one-stop shop for research, planning, and execution. Even better? Only the individual chat is shared — not your entire project or its files/ Projects you can: Each user can create unlimited Projects (with up to 20 files each, subject to subscription rate limits). If you're done with a Project, you can delete it by clicking the three-dot menu next to the Project name. This will permanently erase all chats, files and custom instructions inside. It's a good idea to delete unnecessary chats and files to ensure you never run out of space. Once deleted, the content is purged within 30 days — so be sure to back up anything important. If you have memory enabled on your ChatGPT account, Projects can reference past chats and uploaded files to provide more relevant, consistent answers. This is especially useful for long-term or multi-phase work, like writing a novel or managing a product launch. Note: To enable full memory functionality in Projects, make sure both Saved Memories and Reference Chat History are turned on in your settings. Use the Search chats bar to quickly pull up any conversation across your Projects. You can remove files, merge documents, or break your work into multiple Projects if you hit your file limit. For Pro, Team, Enterprise, and Edu users, ChatGPT does not use project data to train its models. If you're a free user or on a personal Plus/Pro plan, training can happen only if you've opted into model improvement. For enterprise-level users, Projects inherit all your workspace's existing settings — including encryption, audit logs, feature availability and data residency. Admins can't yet disable Projects entirely, but they maintain full control over retention windows and tool access. Whether you're managing a solo side hustle or leading a team initiative, ChatGPT Projects make it easier to keep everything aligned and all in one place. The feature's mix of organization, chat tools and deep memory integration turns ChatGPT into something so much more than a chatbot. It becomes your creative, analytical, always-on partner. It's completely changed the way I work and stay organized.
