Elon Musk has filed a lawsuit accusing OpenAI and its chief executive, Sam Altman, of breaching its foundational mission by putting the pursuit of profit ahead of the benefit of humanity. The world's richest man, a founding board member of the artificial intelligence company behind ChatGPT, claimed that Altman had "set aflame" the OpenAI's founding agreement by signing an investment deal with Microsoft. The lawsuit, filed in San Francisco on Thursday, claims that OpenAI is now developing artificial general intelligence (AGI) – a theoretical form of AI that can perform a range of tasks at or above a human level of intelligence – for profit rather than for the benefit of humankind. "OpenAI Inc has been transformed into a closed-source, de facto subsidiary of the largest technology company in the world: Microsoft. Under its new board, it is not just developing but is actually refining an AGI to maximise profits for Microsoft, rather than for the benefit of humanity," the lawsuit alleges.

Elon Musk sued OpenAI and its Chief Executive Officer Sam Altman, alleging they violated the artificial intelligence startup's founding mission by putting profit ahead of benefiting humanity. The 52-year-old billionaire, who was a co-founder of OpenAI but no longer has a stake, said in a lawsuit filed late Thursday in San Francisco that the company's close relationship with Microsoft Corp. has undermined its original mission of creating open-source technology that wouldn't be subject to corporate priorities. Musk, who is also CEO of Tesla Inc., has been among the most outspoken about the dangers of AI and artificial general intelligence, or AGI. The release of OpenAI's ChatGPT more than a year ago popularized advances in AI technology and raised concerns about the risks surrounding the race to develop AGI, where computers are as smart as an average human. "To this day, OpenAI Inc.'s website continues to profess that its charter is to ensure that AGI'benefits all of humanity,'" the lawsuit said.

One scorching day this past September, I made the dangerous decision to try to circumnavigate some data centers. The ones I chose sit between a regional airport and some farm fields in Goodyear, Arizona, half an hour's drive west of downtown Phoenix. When my Uber pulled up beside the unmarked buildings, the temperature was 97 degrees Fahrenheit. The air crackled with a latent energy, and some kind of pulsating sound was emanating from the electric wires above my head, or maybe from the buildings themselves. With no shelter from the blinding sunlight, I began to lose my sense of what was real. Microsoft announced its plans for this location, and two others not so far away, back in 2019--a week after the company revealed its initial 1 billion investment in OpenAI, the buzzy start-up that would later release ChatGPT.

As generative AI systems like OpenAI's ChatGPT and Google's Gemini become more advanced, they are increasingly being put to work. Startups and tech companies are building AI agents and ecosystems on top of the systems that can complete boring chores for you: think automatically making calendar bookings and potentially buying products. But as the tools are given more freedom, it also increases the potential ways they can be attacked. Now, in a demonstration of the risks of connected, autonomous AI ecosystems, a group of researchers have created one of what they claim are the first generative AI worms--which can spread from one system to another, potentially stealing data or deploying malware in the process. "It basically means that now you have the ability to conduct or to perform a new kind of cyberattack that hasn't been seen before," says Ben Nassi, a Cornell Tech researcher behind the research.

OpenAI has been without a full board since November, after several members of its previous board stepped down amid a chaotic leadership turmoil that started when its CEO, Sam Altman, was fired. OpenAI's previous board fired Altman saying he had not been "consistently candid in his communications" with board members. After several days of furious backlash from employees and investors, Altman was reinstated as CEO and the company announced a new council meant to set up the new board.

Hundreds of millions of people now interact with language models, with uses ranging from serving as a writing aid to informing hiring decisions. Yet these language models are known to perpetuate systematic racial prejudices, making their judgments biased in problematic ways about groups like African Americans. While prior research has focused on overt racism in language models, social scientists have argued that racism with a more subtle character has developed over time. It is unknown whether this covert racism manifests in language models. Here, we demonstrate that language models embody covert racism in the form of dialect prejudice: we extend research showing that Americans hold raciolinguistic stereotypes about speakers of African American English and find that language models have the same prejudice, exhibiting covert stereotypes that are more negative than any human stereotypes about African Americans ever experimentally recorded, although closest to the ones from before the civil rights movement. By contrast, the language models' overt stereotypes about African Americans are much more positive. We demonstrate that dialect prejudice has the potential for harmful consequences by asking language models to make hypothetical decisions about people, based only on how they speak. Language models are more likely to suggest that speakers of African American English be assigned less prestigious jobs, be convicted of crimes, and be sentenced to death. Finally, we show that existing methods for alleviating racial bias in language models such as human feedback training do not mitigate the dialect prejudice, but can exacerbate the discrepancy between covert and overt stereotypes, by teaching language models to superficially conceal the racism that they maintain on a deeper level. Our findings have far-reaching implications for the fair and safe employment of language technology.

In the rapidly advancing arena of large language models (LLMs), a key challenge is to enhance their capabilities amid a looming shortage of high-quality training data. Our study starts from an empirical strategy for the light continual training of LLMs using their original pre-training data sets, with a specific focus on selective retention of samples that incur moderately high losses. These samples are deemed informative and beneficial for model refinement, contrasting with the highest-loss samples, which would be discarded due to their correlation with data noise and complexity. We then formalize this strategy into a principled framework of Instance-Reweighted Distributionally Robust Optimization (IR-DRO). IR-DRO is designed to dynamically prioritize the training focus on informative samples through an instance reweighting mechanism, streamlined by a closed-form solution for straightforward integration into established training protocols. Through rigorous experimentation with various models and datasets, our findings indicate that our sample-targeted methods significantly improve LLM performance across multiple benchmarks, in both continual pre-training and instruction tuning scenarios. Our codes are available at https://github.com/VITA-Group/HardFocusTraining.

As LLMs become ubiquitous and integrated into numerous digital applications, it's an increasingly pressing research problem to understand the internal mechanisms that underlie their behaviour - this is the problem of mechanistic interpretability. A fundamental subproblem is to causally attribute particular behaviours to individual parts of the transformer forward pass, corresponding to specific components (such as attention heads, neurons, layer contributions, or residual streams), often at specific positions in the input token sequence. This is important because in numerous case studies of complex behaviours, they are found to be driven by sparse subgraphs within the model (Meng et al., 2023; Olsson et al., 2022; Wang et al., 2022). A classic form of causal attribution uses zero-ablation, or knock-out, where a component is deleted and we see if this negatively affects a model's output - a negative effect implies the component was causally important. More recent work has generalised this to replacing a component's activations with samples from some baseline distribution (with zero-ablation being a special case where activations are resampled to be zero). We focus on the popular and widely used method of Activation Patching (also known as causal mediation analysis) (Chan et al., 2022; Geiger et al., 2022; Meng et al., 2023) where the baseline distribution is a component's activations on some corrupted input, such as an alternate string with a different answer (Pearl, 2001; Robins and Greenland, 1992). Given a causal attribution method, it is common to sweep across all model components, directly evaluating the effect of intervening on each of them via resampling (Meng et al., 2023). However, when working with SoTA models it can be expensive to attribute behaviour especially to small components (e.g.

The safety alignment of Large Language Models (LLMs) is vulnerable to both manual and automated jailbreak attacks, which adversarially trigger LLMs to output harmful content. However, current methods for jailbreaking LLMs, which nest entire harmful prompts, are not effective at concealing malicious intent and can be easily identified and rejected by well-aligned LLMs. This paper discovers that decomposing a malicious prompt into separated sub-prompts can effectively obscure its underlying malicious intent by presenting it in a fragmented, less detectable form, thereby addressing these limitations. We introduce an automatic prompt \textbf{D}ecomposition and \textbf{R}econstruction framework for jailbreak \textbf{Attack} (DrAttack). DrAttack includes three key components: (a) `Decomposition' of the original prompt into sub-prompts, (b) `Reconstruction' of these sub-prompts implicitly by in-context learning with semantically similar but harmless reassembling demo, and (c) a `Synonym Search' of sub-prompts, aiming to find sub-prompts' synonyms that maintain the original intent while jailbreaking LLMs. An extensive empirical study across multiple open-source and closed-source LLMs demonstrates that, with a significantly reduced number of queries, DrAttack obtains a substantial gain of success rate over prior SOTA prompt-only attackers. Notably, the success rate of 78.0\% on GPT-4 with merely 15 queries surpassed previous art by 33.1\%. The project is available at https://github.com/xirui-li/DrAttack.

Generative AI has the potential to create a new form of interactive media: AI-bridged creative language arts (CLA), which bridge the author and audience by personalizing the author's vision to the audience's context and taste at scale. However, it is unclear what the authors' values and attitudes would be regarding AI-bridged CLA. To identify these values and attitudes, we conducted an interview study with 18 authors across eight genres (e.g., poetry, comics) by presenting speculative but realistic AI-bridged CLA scenarios. We identified three benefits derived from the dynamics between author, artifact, and audience: those that 1) authors get from the process, 2) audiences get from the artifact, and 3) authors get from the audience. We found how AI-bridged CLA would either promote or reduce these benefits, along with authors' concerns. We hope our investigation hints at how AI can provide intriguing experiences to CLA audiences while promoting authors' values.