An official website of the United States government
Abstract PurposeThe COVID-19 pandemic and associated public health control measures resulted in both higher rates of intimate partner abuse and violence (IPA/V) and more severe victimization. Domestic violence advocacy programs struggled to maintain organizational capacity to provide survivor-centered services in the face of both increased demand and rapid changes necessary to mitigate disease spread. The current study explores ways that legal advocates and the legal systems responded to the needs of IPA/V survivors. MethodsLeaders of 25 state and territory Coalitions across the U.S. participated in the study. Semi-structured interview questions were based on rapidly emerging areas of concern and drew on possible strengths and weaknesses in direct service provision during the height of the COVID-19 pandemic, including questions about public health control measures, service provision, gaps in services, and lessons learned. ResultsInterviews with Coalition leaders revealed gaps in legal system responses during the pandemic, but also suggested new directions for service delivery. Four main themes emerged: lack of access to the legal system, limitations of in-person legal system responses, limitations of virtual legal system responses, and changes needed moving forward. ConclusionBacklogs in case processing communicate to survivors and the larger community that responding to IPA/V is not urgent. Advocates faced difficulty supporting survivors in person while virtual hearings sometimes presented other challenges for advocacy. However, some changes, including innovative online services and broad resolve to center BIPOC survivor voices, have the potential to enhance safety for survivors and push the movement forward.
more »
« less
Authorbots
ChatGPT has exploded into the popular consciousness in recent months, and the hype and concerns about the program have only grown louder with the release of GPT-4, a more powerful version of the software. Its deployment, including with applications such as Microsoft Office, has raised questions about whether the developers or distributors of code that includes ChatGPT, or similar generative pre-trained transformers, could face liability for tort claims such as defamation or false light. One important potential barrier to these claims is the immunity con-ferred by 47 U.S.C. § 230, popularly known as “Section 230.” In this Essay, we make two claims. First, Section 230 is likely to protect the creators, distributors, and hosts of online services that include ChatGPT in many cases. Users of those services, though, may be at greater legal risk than is commonly believed. Second, ChatGPT and its ilk make the analysis of the Section 230 safe harbor more com-plex, both substantively and procedurally. This is likely a negative consequence for the software’s developers and hosts, since complexity in law tends to generate uncertainty, which in turn creates cost. Nonetheless, we contend that Section 230 has more of a role to play in legal questions about ChatGPT than most commentators do—including the principal legislative drafters of Section 230—and that this result is generally a desirable one.
more »
« less
- Award ID(s):
- 2217215
- PAR ID:
- 10488665
- Editor(s):
- Volokh, Eugene
- Publisher / Repository:
- Journal of Free Speech Law
- Date Published:
- Journal Name:
- Journal of Free Speech Law
- Volume:
- 3
- Issue:
- 2
- ISSN:
- 2766-7677
- Subject(s) / Keyword(s):
- AI, algorithm, artificial intelligence, ChatGPT, defamation, liability, libel, machine learning, OpenAI, Section 230, software, tort
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Regulations outline high-level guidance or expectations for a profession or industry. Analyzing laws or regulations is one way a software developer would derive and document regulatory compliance requirements within their software design. However, ambiguities within regulations can make it challenging to define technical software design specifications for regulatory requirements. Further, due to the subjective nature of ambiguous phrasing within a law or regulation, the interpretation of the legal text can differ based on the interpreter’s perspective. Our study examines whether software developers can analyze regulatory ambiguities as a group using our modeling process and our online Ambiguity Heuristics Analysis Builder (AHAB) tool. Eleven participants formed three groups and modeled ambiguities within a regulation using our process and tool. Modeling regulatory ambiguity, while difficult for our participants, allowed them to communicate potential issues, ask meaningful questions, and deepen their knowledge of the regulation. Ambiguity modeling allows developers to articulate interpretation and compliance issues with the laws to other parties (i.e., lawyers) and document this requirement analysis step for future use. Documenting these intermediate steps is rarely highlighted in requirement analysis. However, it is useful to negotiate with regulators, avoid negligence, and show due diligence toward regulatory compliance. It can also lead to clarifying guidance software developers need to make better, more compliant choices during software design.more » « less
-
While the United States currently has no comprehensive privacy law, the Children’s Online Privacy Protection Act (“COPPA”) has been in effect for over twenty years. As a result, the study of compliance issues among child-directed online services can yield important lessons for future enforcement efforts and can be used to inform the design of future state and federal privacy laws designed to protect people of all ages. This Essay describes relevant research conducted to understand privacy compliance issues and how that has led the author to several recommendations for how privacy enforcement can be improved more generally. While these recommendations are informed by the study of child-directed services’ compliance with COPPA, they are applicable to future state and federal privacy laws aimed at protecting the general public (i.e., not just children). Despite evidence of thousands of COPPA violations (e.g., one study found evidence that a majority of child-directed mo-bile apps appeared to be violating COPPA in various ways), the Federal Trade Commission (“FTC”) and state attorneys general — the only entities with enforcement authority under the law — pursue few enforcement efforts each year. Despite having competent personnel, these organizations are heavily constrained and under-resourced — as a result, enforcement by regulators is simply not seen as a credible threat by software developers. Research has found that developers are much more concerned with apps being removed from app stores (i.e., due to enforcement of platforms’ terms of service) than with the largely theoretical threat of regulatory enforcement. Yet the burden of COPPA compliance largely rests on numerous individual app developers. Thus, shifting enforcement efforts to the far-fewer platforms that distribute the apps (and make representations about their privacy and security properties) and data recipients (who ultimately receive consumers’ identifiable data) would likely yield better outcomes for consumers, while allowing the FTC to better focus its enforcement efforts and have greater impact. Based on these observations, this Essay proposes a new enforcement framework. In this framework, compliance burdens are shifted away from the numerous individual online services to the fewer bigger players who are best positioned to comply: platforms and third-party data recipients. The FTC’s limited resources can then focus on those entities at the top of the data food chain. Enforcement targeting the other, more numerous, individual online services could be left to a novel mechanism that uses a private right of action to foster more robust industry self-regulation through FTC-approved certification programs.more » « less
-
A major risk of using language models in practical applications is their tendency to hallucinate incorrect statements. Hallucinations are often attributed to knowledge gaps in LMs, but we hypothesize that in some cases, when justifying previously generated hallucinations, LMs output false claims that they can separately recognize as incorrect. We construct three question-answering datasets where ChatGPT and GPT-4 often state an incorrect answer and offer an explanation with at least one incorrect claim. Crucially, we find that ChatGPT and GPT-4 can identify 67% and 87% of their own mistakes, respectively. We refer to this phenomenon as hallucination snowballing: an LM over-commits to early mistakes, leading to more mistakes that it otherwise would not make.more » « less
-
null (Ed.)Background In the 1970s, the Federal Trade Commission declared that allowing medical providers to advertise directly to consumers would be “providing the public with truthful information about the price, quality or other aspects of their service.” However, our understanding of the advertising content is highly limited. Objective To assess whether direct-to-consumer medical service advertisements provide relevant information on access, quality and cost of care, a content analysis was conducted. Method Television and online advertisements for medical services directly targeting consumers were collected in two major urban centres in Nevada, USA, identifying 313 television advertisements and 200 non-duplicate online advertisements. Results Both television and online advertisements reliably conveyed information about the services provided and how to make an appointment. At the same time, less than half of the advertisements featured insurance information and hours of operation and less than a quarter of them contained information regarding the quality and price of care. The claims of quality were substantiated in even fewer advertisements. The scarcity of quality and cost information was more severe in television advertisements. Conclusion There is little evidence that medical service advertising, in its current form, would contribute to lower prices or improved quality of care by providing valuable information to consumers.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- The DOI is not currently available.
