In August of 2024, 495 hackers generated evaluations in an open-ended bug bounty targeting the Open Language Model (OLMo) from The Allen Institute for AI. A vendor panel staffed by representatives of OLMo's safety program adjudicated changes to OLMo's documentation and awarded cash bounties to participants who successfully demonstrated a need for public disclosure clarifying the intent, capacities, and hazards of model deployment. This paper presents a collection of lessons learned, illustrative of flaw reporting best practices intended to reduce the likelihood of incidents and produce safer large language models (LLMs). These include best practices for safety reporting processes, their artifacts, and safety program staffing.

As artificial intelligence (AI) systems become increasingly deployed across the world, they are also increasingly implicated in AI incidents – harm events to individuals and society. As a result, industry, civil society, and governments worldwide are developing best practices and regulations for monitoring and analyzing AI incidents. The AI Incident Database (AIID) is a project that catalogs AI incidents and supports further research by providing a platform to classify incidents for different operational and research-oriented goals. This study reviews the AIID’s dataset of 750+ AI incidents and two independent taxonomies applied to these incidents to identify common challenges to indexing and analyzing AI incidents. We find that certain patterns of AI incidents present structural ambiguities that challenge incident databasing and explore how epistemic uncertainty in AI incident reporting is unavoidable. We therefore report mitigations to make incident processes more robust to uncertainty related to cause, extent of harm, severity, or technical details of implicated systems. With these findings, we discuss how to develop future AI incident reporting practices.

Large language model (LLM) benchmarks inform LLM use decisions (e.g., "is this LLM safe to deploy for my use case and context?"). However, benchmarks may be rendered unreliable by various failure modes impacting benchmark bias, variance, coverage, or people's capacity to understand benchmark evidence. Using the National Institute of Standards and Technology's risk management process as a foundation, this research iteratively analyzed 26 popular benchmarks, identifying 57 potential failure modes and 196 corresponding mitigation strategies. The mitigations reduce failure likelihood and/or severity, providing a frame for evaluating "benchmark risk," which is scored to provide a metaevaluation benchmark: BenchRisk. Higher scores indicate benchmark users are less likely to reach an incorrect or unsupported conclusion about an LLM. All 26 scored benchmarks present significant risk within one or more of the five scored dimensions (comprehensiveness, intelligibility, consistency, correctness, and longevity), which points to important open research directions for the field of LLM benchmarking. The BenchRisk workflow allows for comparison between benchmarks; as an open-source tool, it also facilitates the identification and sharing of risks and their mitigations.

Intelligent system audits are labor-intensive assurance activities that are typically performed once and discarded along with the opportunity to programmatically test all similar products for the market. This study illustrates how several incidents (i.e., harms) involving Named Entity Recognition (NER) can be prevented by scaling up a previously-performed audit of NER systems. The audit instrument's diagnostic capacity is maintained through a security model that protects the underlying data (i.e., addresses Goodhart's Law). An open-source evaluation infrastructure is released along with an example derived from a real-world audit that reports aggregated findings without exposing the underlying data.

Widespread adoption of artificial intelligence (AI) technologies is substantially affecting the human condition in ways that are not yet well understood. Negative unintended consequences abound including the perpetuation and exacerbation of societal inequalities and divisions via algorithmic decision making. We present six grand challenges for the scientific community to create AI technologies that are human-centered, that is, ethical, fair, and enhance the human condition. These grand challenges are the result of an international collaboration across academia, industry and government and represent the consensus views of a group of 26 experts in the field of human-centered artificial intelligence (HCAI). In essence, these challenges advocate for a human-centered approach to AI that (1) is centered in human well-being, (2) is designed responsibly, (3) respects privacy, (4) follows human-centered design principles, (5) is subject to appropriate governance and oversight, and (6) interacts with individuals while respecting human’s cognitive capacities. We hope that these challenges and their associated research directions serve as a call for action to conduct research and development in AI that serves as a force multiplier towards more fair, equitable and sustainable societies.

Mature industrial sectors (e.g., aviation) collect their real world failures in incident databases to inform safety improvements. Intelligent systems currently cause real world harms without a collective memory of their failings. As a result, companies repeatedly make the same mistakes in the design, development, and deployment of intelligent systems. A collection of intelligent system failures experienced in the real world (i.e., incidents) is needed to ensure intelligent systems benefit people and society. The AI Incident Database is an incident collection initiated by an industrial/non-profit cooperative to enable AI incident avoidance and mitigation. The database supports a variety of research and development use cases with faceted and full text search on more than 1,000 incident reports archived to date.

Synthetic media detection technologies label media as either synthetic or non-synthetic and are increasingly used by journalists, web platforms, and the general public to identify misinformation and other forms of problematic content. As both well-resourced organizations and the non-technical general public generate more sophisticated synthetic media, the capacity for purveyors of problematic content to adapt induces a detection dilemma : as detection practices become more accessible, they become more easily circumvented. This paper describes how a multistakeholder cohort from academia, technology platforms, media entities, and civil society organizations active in synthetic media detection and its socio-technical implications evaluates the detection dilemma. Specifically, we offer an assessment of detection contexts and adversary capacities sourced from the broader, global AI and media integrity community concerned with mitigating the spread of harmful synthetic media. A collection of personas illustrates the intersection between unsophisticated and highly-resourced sponsors of misinformation in the context of their technical capacities. This work concludes that there is no "best'' approach to navigating the detector dilemma, but derives a set of implications from multistakeholder input to better inform detection process decisions and policies, in practice.

Researchers in AI and Operations Research employ the framework of Markov Decision Processes (MDPs) to formalize problems of sequential decision making under uncertainty. A common approach is to implement a simulator of the stochastic dynamics of the MDP and a Monte Carlo optimization algorithm that invokes this simulator to solve the MDP. The resulting software system is often realized by integrating several systems and functions that are collectively subject to failures of specification, implementation, integration, and optimization. We present these failures as queries for a computational steering visual analytic system (MDPVIS). MDPVIS addresses three visualization research gaps. First, the data acquisition gap is addressed through a general simulator-visualization interface. Second, the data analysis gap is addressed through a generalized MDP information visualization. Finally, the cognition gap is addressed by exposing model components to the user. MDPVIS generalizes a visualization for wildfire management. We use that problem to illustrate MDPVIS.

Markov Decision Process (MDP) simulators and optimization algorithms integrate several systems and functions that are collectively subject to failures of specification, implementation, integration, and optimization. We present a domain agnostic visual analytic design and implementation for testing and debugging MDPs: MDPVIS.