Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-1560
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-10-16T10:41:32.720178Z
status: published
lens: p_space
sector: health
headline: Using Artificial Intelligence to Access Reproductive Healthcare Information by U.S. States Following Sexual Assault
dek: There is currently wide variability across U.S. states regarding the legality of abortion with some states having relatively open access to abortion services and others with a total or near total ban on abortion access. Further complicating these state-by-state differences is the variability in policies and procedures with regard to exemptions (e.g., rape, incest) that may allow women access to abortion services that otherwise would be banned in that state. Factors such as a victim's alcohol use at the time of r…
gain_title: (none)
problem_title: In responses to prompts from women pregnant after rape, ChatGPT-4o often omitted helpful information and empathic wording, which may worsen existing disparities between U.S. states.
trace_subject: (none)
gain_reading: (none)
gain_evidence: (none)
problem_reading: In responses to prompts from women pregnant after rape, ChatGPT-4o often omitted helpful information and empathic wording, which may worsen existing disparities between U.S. states.
problem_evidence: responses often failed to include helpful information and statements reflective of empathic tone | which may exacerbate pre-existing disparities across U.S. states
quick_read: Researchers tested how ChatGPT-4o answered questions about getting an abortion after becoming pregnant from rape. They ran four prompts, varying by alcohol use and by whether the user acknowledged the rape, in each of the 50 U.S. states. Data were collected between October 31 and November 11, 2024, and responses were coded for accuracy, helpfulness and empathy.

The findings matter because state abortion laws and rape exemptions vary widely, and women may turn to AI tools to navigate that confusion. Responses did not differ meaningfully by prompt type, but they often lacked helpful content and empathic tone. The abstract does not report how users acted on the answers or what outcomes followed.
limitation: The study examined one model, ChatGPT-4o, using prompts collected in a short window in late 2024, so results may not apply to other tools or later model versions.
tag: Evidence-backed problem
key_points: The study analyzed ChatGPT-4o responses across all 50 U.S. states using four prompt variations. | Prompts varied by alcohol use at the time of rape and by rape acknowledgment. | Responses were qualitatively coded for accuracy, helpfulness and empathic statements. | States with more severe abortion restrictions were more likely to get information on rape exemptions. | No significant differences emerged across the prompt conditions.
rundown: The paper starts from the fact that abortion legality varies widely across U.S. states, from relatively open access to total or near total bans. Exemption rules, such as those for rape or incest, add further variation that can confuse survivors seeking care.

The authors note that alcohol use during an assault and a victim's rape acknowledgment may delay or prevent medical care. They therefore built prompts around those two factors to see whether the chatbot's answers changed.

The main differences found were by state rather than by prompt. Responses for states with stricter restrictions more often mentioned rape exemptions, but responses overall often lacked helpful information and empathic tone.
sources:
- peer_reviewed | Archives of Sexual Behavior | https://doi.org/10.1007/s10508-026-03608-w | 2026-10-15
prev: 0000000000000000000000000000000000000000000000000000000000000000

sha256

07224c0d6839e4884578e441ac733efe65d3dbaeffeae76705cd1b89c7e54713

previous

0000000000000000000000000000000000000000000000000000000000000000