The news on Friday that Sam Altman, the chief executive of OpenAI, had been abruptly sacked by the company's board came as a shock to the tech industry. "Mr Altman's departure," said the ponderous announcement, "follows a deliberative review process by the board, which concluded that he was not consistently candid in his communications with the board, hindering its ability to exercise its responsibilities. The board no longer has confidence in his ability to continue leading OpenAI." Given that, ever since ChatGPT took the world by storm last December, Altman has been the photogenic poster-boy for generative AI – the darling of the mainstream media and an honoured invitee to the corridors of western power – news of his sudden fall from grace launched a torrent of excited speculation in the tech commentariat. Nobody, it seems, actually knew anything, but there was a consensus that Something Was Up. No doubt we will get to the bottom of the mystery in due course, but for now a more productive line of inquiry might be into the corporate history of OpenAI.

Following his surprise firing on Friday, currently-former OpenAI CEO Sam Altman might not be as out of a job as we initially thought he was, according to new word from The Verge on Saturday. Reportedly, sources close to Altman say that the board itself, in a stunning reversal, have "agreed in principal" to resign while reinstating him to his former position. However, the board has since reportedly missed a 5pm PT deadline regarding the decision. Shortly after Altman's firing on Friday afternoon, several senior staffers including former Chairman and President Greg Brockman, Director of Research Jakub Pachocki, Head of Preparedness Aleksander Madry and Senior Researcher Szymon Sidor tendered their resignations in protest. Numerous additional OpenAI staffers were set to quit in solidarity at that meeting.

Two of the board members who voted Altman out worked for think tanks backed by Open Philanthropy, a tech billionaire-backed foundation that supports projects preventing AI from causing catastrophic risk to humanity: Helen Toner, the director of strategy and foundational research grants for Center for Security and Emerging Technology at Georgetown, and Tasha McCauley, whose LinkedIn profile says she began work as an adjunct senior management scientist at Rand Corporation earlier this year. Toner has previously spoken at conferences for a philanthropic movement closely tied to AI safety. McCauley is also involved in the work.

OpenAI investors are pressing the company's board to reverse its decision to fire Sam Altman as chief executive officer and remove him as a director, according to people with knowledge of the matter. Some of the investors including Thrive Global are also in talks with Microsoft, the largest shareholder of OpenAI, said the people, who asked not to be identified because the overtures are private. Altman is open to returning to the company, one of the people said. In one scenario under consideration, members of the current OpenAI board would step down as soon as this weekend, according to multiple with knowledge of the situation. However, the situation is still fluid and no decisions have been made, the people said.

For most of his life, Sam Altman has resembled an accelerating train. He founded a startup as a teenager, achieved middling success -- and then jumped to running Silicon Valley's premier startup accelerator. After that, Altman co-founded OpenAI, whose ChatGPT tool has whipped up a frenzy of excitement about generative AI. Now 38, Altman has become the face of an AI-fueled future, traveling around the world to explain to world leaders and everyone else how the technology he'd helped create would change everything about human existence. He was the most Silicon Valley person alive.

We introduce a large language model (LLM) based approach to answer complex questions requiring multi-hop numerical reasoning over financial reports. While LLMs have exhibited remarkable performance on various natural language and reasoning tasks, complex reasoning problems often rely on few-shot prompts that require carefully crafted examples. In contrast, our approach uses novel zero-shot prompts that guide the LLM to encode the required reasoning into a Python program or a domain specific language. The generated program is then executed by a program interpreter, thus mitigating the limitations of LLM in performing accurate arithmetic calculations. We evaluate the proposed approach on three financial datasets using some of the recently developed generative pretrained transformer (GPT) models and perform comparisons with various zero-shot baselines. The experimental results demonstrate that our approach significantly improves the accuracy for all the LLMs over their respective baselines. We provide a detailed analysis of the results, generating insights to support our findings. The success of our approach demonstrates the enormous potential to extract complex domain specific numerical reasoning by designing zero-shot prompts to effectively exploit the knowledge embedded in LLMs.

We explore developing rich semantic models of systems. Specifically, we consider structured causal explanations about state changes in those systems. Essentially, we are developing process-based dynamic knowledge graphs. As an example, we construct a model of the causal threads for geological changes proposed by the Snowball Earth theory. Further, we describe an early prototype of a graphical interface to present the explanations. Unlike statistical approaches to summarization and explanation such as Large Language Models (LLMs), our approach of direct representation can be inspected and verified directly.

While Transformers have shown remarkable success in natural language processing, their attention mechanism's large memory requirements have limited their ability to handle longer contexts. Prior approaches, such as recurrent memory or retrieval-based augmentation, have either compromised the random-access flexibility of attention (i.e., the capability to select any token in the entire context) or relied on separate mechanisms for relevant context retrieval, which may not be compatible with the model's attention. In this paper, we present a novel approach that allows access to the complete context while retaining random-access flexibility, closely resembling running attention on the entire context. Our method uses a landmark token to represent each block of the input and trains the attention to use it for selecting relevant blocks, enabling retrieval of blocks directly through the attention mechanism instead of by relying on a separate mechanism. Our approach seamlessly integrates with specialized data structures and the system's memory hierarchy, enabling processing of arbitrarily long context lengths. We demonstrate that our method can obtain comparable performance with Transformer-XL while significantly reducing the number of retrieved tokens in each step. Finally, we show that fine-tuning LLaMA 7B with our method successfully extends its context length capacity to over 32k tokens, allowing for inference at the context lengths of GPT-4. We release the implementation of landmark attention and the code to reproduce our experiments at https://github.com/epfml/landmark-attention/.

Numerous studies have proved their effective strength in detecting Control Area Network (CAN) attacks. In the realm of understanding the human semantic space, transformer-based models have demonstrated remarkable effectiveness. Leveraging pre-trained transformers has become a common strategy in various language-related tasks, enabling these models to grasp human semantics more comprehensively. To delve into the adaptability evaluation on pre-trained models for CAN intrusion detection, we have developed two distinct models: CAN-SecureBERT and CAN-LLAMA2. Notably, our CAN-LLAMA2 model surpasses the state-of-the-art models by achieving an exceptional performance 0.999993 in terms of balanced accuracy, precision detection rate, F1 score, and a remarkably low false alarm rate of 3.10e-6. Impressively, the false alarm rate is 52 times smaller than that of the leading model, MTH-IDS (Multitiered Hybrid Intrusion Detection System). Our study underscores the promise of employing a Large Language Model as the foundational model, while incorporating adapters for other cybersecurity-related tasks and maintaining the model's inherent language-related capabilities.

In the rapidly evolving landscape of artificial intelligence, ChatGPT has been widely used in various applications. The new feature -- customization of ChatGPT models by users to cater to specific needs has opened new frontiers in AI utility. However, this study reveals a significant security vulnerability inherent in these user-customized GPTs: prompt injection attacks. Through comprehensive testing of over 200 user-designed GPT models via adversarial prompts, we demonstrate that these systems are susceptible to prompt injections. Through prompt injection, an adversary can not only extract the customized system prompts but also access the uploaded files. This paper provides a first-hand analysis of the prompt injection, alongside the evaluation of the possible mitigation of such attacks. Our findings underscore the urgent need for robust security frameworks in the design and deployment of customizable GPT models. The intent of this paper is to raise awareness and prompt action in the AI community, ensuring that the benefits of GPT customization do not come at the cost of compromised security and privacy.