{"id":"5adcf12ea176ab2a92d12d7eb1bae5e4be5e5ef9eb0bae09be3b55ec447c8b04","title":"MIT Technology Review argues GPT-5’s health push raises accountability questions","argument":"MIT Technology Review published this article on August 12, 2025. It says GPT-5 has mostly underwhelmed: the release is more a product update than a breakthrough, with users highlighting mistakes and problems with the model’s automatic choice of which system to use. The article also notes that GPT-5 seems less likely to flatter users excessively.\n\nIts larger concern is OpenAI’s shift from promoting broad AI capabilities to encouraging particular uses, especially health advice. The publication describes OpenAI presenting a couple’s account of using ChatGPT to understand cancer diagnoses and consider treatment decisions. It argues that evidence about doctors using AI as a clinical tool does not establish that people without medical backgrounds should seek personal health advice from AI. It also points to a medical-journal report about a man who nearly died after a conversation with ChatGPT.\n\nThe article frames the unresolved stake as accountability. It says there is little indication that technology companies will be liable for harm caused by AI health advice, and leaves open what recourse users would have when harmful advice is given.","topicIds":["artificial-general-intelligence"],"domain":"other","claimType":"attributed reading","limits":"Text as retrieved on November 1, 2026; originally published August 12, 2025. The article does not establish how legal responsibility for harmful AI health advice would be determined. The quoted question about recourse is raised by Damien Williams.","speaker":{"name":"MIT Technology Review","roleAtSource":"Publisher"},"source":{"title":"What you may have missed about GPT-5","publisher":"MIT Technology Review","format":"Article","url":"https://www.technologyreview.com/2025/08/12/1121565/what-you-may-have-missed-about-gpt-5/","publishedAt":"2025-08-12T07:52:23Z","dateBasis":"Date from the publisher’s own records, through its API."},"evidence":[{"text":"MIT Technology Review characterizes GPT-5’s reception as mostly underwhelming and the release as a product update rather than a breakthrough.","excerpt":"Overall, as my colleague Grace Huckins pointed out, the new release represents more of a product update","url":"https://www.technologyreview.com/2025/08/12/1121565/what-you-may-have-missed-about-gpt-5/","relationship":"Source-stated"},{"text":"The publication argues that AI companies are increasingly promoting specific applications, and suggests one possible reason is a lack of expected technical breakthroughs.","excerpt":"A possible explanation for this pivot is that tech giants simply have not made the breakthroughs they’ve expected.","url":"https://www.technologyreview.com/2025/08/12/1121565/what-you-may-have-missed-about-gpt-5/","relationship":"Source-stated"},{"text":"The article says OpenAI began explicitly encouraging people to use its models for health advice with GPT-5’s launch.","excerpt":"With the launch of GPT-5, OpenAI has begun explicitly telling people to use its models for health advice.","url":"https://www.technologyreview.com/2025/08/12/1121565/what-you-may-have-missed-about-gpt-5/","relationship":"Source-stated"},{"text":"The publication argues that evidence about doctors benefiting from AI as a clinical tool does not establish that people without medical backgrounds should use AI for personal health advice.","excerpt":"it’s using evidence that doctors can benefit from AI as a clinical tool, as in the Kenya study, to suggest that people without any medical background should ask the AI model for advice about their own health.","url":"https://www.technologyreview.com/2025/08/12/1121565/what-you-may-have-missed-about-gpt-5/","relationship":"Source-stated"},{"text":"The article reports that a medical journal published a paper about a man who nearly died after a conversation with ChatGPT.","excerpt":"the Annals of Internal Medicine published a paper about a man who stopped eating salt and began ingesting dangerous amounts of bromide following a conversation with ChatGPT.","url":"https://www.technologyreview.com/2025/08/12/1121565/what-you-may-have-missed-about-gpt-5/","relationship":"Source-stated"},{"text":"The publication frames accountability and users’ recourse for harmful AI health advice as unresolved.","excerpt":"When ChatGPT gives you harmful medical advice because it’s been trained on prejudicial data, or because ‘hallucinations’ are inherent in the operations of the system, what’s your recourse?","url":"https://www.technologyreview.com/2025/08/12/1121565/what-you-may-have-missed-about-gpt-5/","relationship":"Source-stated"}],"sourceMedia":{"status":"unchecked","sourceUrl":"https://www.technologyreview.com/2025/08/12/1121565/what-you-may-have-missed-about-gpt-5/","checkedAt":null,"items":[]},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"When chatbot health advice causes harm, what recourse will users have, and who would be held responsible?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"MIT Technology Review argues GPT-5’s health push raises accountability questions","summary":"MIT Technology Review published this article on August 12, 2025. It says GPT-5 has mostly underwhelmed: the release is more a product update than a breakthrough, with users highlighting mistakes and problems with the model’s automatic choice of which system to use. The article also notes that GPT-5 seems less likely to flatter users excessively.\n\nIts larger concern is OpenAI’s shift from promoting broad AI capabilities to encouraging particular uses, especially health advice. The publication describes OpenAI presenting a couple’s account of using ChatGPT to understand cancer diagnoses and consider treatment decisions. It argues that evidence about doctors using AI as a clinical tool does not establish that people without medical backgrounds should seek personal health advice from AI. It also points to a medical-journal report about a man who nearly died after a conversation with ChatGPT.\n\nThe article frames the unresolved stake as accountability. It says there is little indication that technology companies will be liable for harm caused by AI health advice, and leaves open what recourse users would have when harmful advice is given.","question":"When chatbot health advice causes harm, what recourse will users have, and who would be held responsible?","claims":[{"text":"MIT Technology Review characterizes GPT-5’s reception as mostly underwhelming and the release as a product update rather than a breakthrough.","excerpt":"Overall, as my colleague Grace Huckins pointed out, the new release represents more of a product update","support":"source-stated"},{"text":"The publication argues that AI companies are increasingly promoting specific applications, and suggests one possible reason is a lack of expected technical breakthroughs.","excerpt":"A possible explanation for this pivot is that tech giants simply have not made the breakthroughs they’ve expected.","support":"source-stated"},{"text":"The article says OpenAI began explicitly encouraging people to use its models for health advice with GPT-5’s launch.","excerpt":"With the launch of GPT-5, OpenAI has begun explicitly telling people to use its models for health advice.","support":"source-stated"},{"text":"The publication argues that evidence about doctors benefiting from AI as a clinical tool does not establish that people without medical backgrounds should use AI for personal health advice.","excerpt":"it’s using evidence that doctors can benefit from AI as a clinical tool, as in the Kenya study, to suggest that people without any medical background should ask the AI model for advice about their own health.","support":"source-stated"},{"text":"The article reports that a medical journal published a paper about a man who nearly died after a conversation with ChatGPT.","excerpt":"the Annals of Internal Medicine published a paper about a man who stopped eating salt and began ingesting dangerous amounts of bromide following a conversation with ChatGPT.","support":"source-stated"},{"text":"The publication frames accountability and users’ recourse for harmful AI health advice as unresolved.","excerpt":"When ChatGPT gives you harmful medical advice because it’s been trained on prejudicial data, or because ‘hallucinations’ are inherent in the operations of the system, what’s your recourse?","support":"source-stated"}],"limitations":["Text as retrieved on November 1, 2026; originally published August 12, 2025.","The article does not establish how legal responsibility for harmful AI health advice would be determined. The quoted question about recourse is raised by Damien Williams."],"mediaId":null,"correctionReason":null,"id":"5adcf12ea176ab2a92d12d7eb1bae5e4be5e5ef9eb0bae09be3b55ec447c8b04","publicationId":"652f3ab1-99f1-cdae-fdbb-329f3a407f68","version":1,"sourceUrl":"https://www.technologyreview.com/2025/08/12/1121565/what-you-may-have-missed-about-gpt-5/","sourceTitle":"What you may have missed about GPT-5","publisher":"MIT Technology Review","publishedAt":"2025-08-12T07:52:23Z","originalPublished":{"value":"2025-08-12T07:52:23Z","precision":"instant","basis":"backfill:api_published"},"firstObservedAt":"2026-11-01T04:33:26.624Z","createdAt":"2026-11-12T15:28:25.506Z","historical":true,"entityIds":[],"topicIds":["artificial-general-intelligence"],"media":[],"sourceMedia":{"status":"unchecked","sourceUrl":"https://www.technologyreview.com/2025/08/12/1121565/what-you-may-have-missed-about-gpt-5/","checkedAt":null,"items":[]},"editorialDecision":{"outcome":"accept","mediaId":null,"inputHash":"f20da6947ace24fbafd3b554df084eb8b77e2efffd9279aa8c0bb93b7f3ec93e","rubricVersion":"unrecorded","provider":"typesafe","model":"jev-1.13.0","reasonCodes":["bridge_admitted_by_model"],"confidence":0.8,"decidedAt":"2026-11-01T04:33:23.819Z"},"provenance":{"textHash":"819aa0cea6973dcfa266d265cb736d06490a7dd381a805f26cda541719faf251","observationId":"bd889d9f70269c06300f273e94f6a165a1f326254f7c57d1500fccde33e51708","retrievedAt":"2026-11-01T04:35:20.190Z","kind":"publisher-page","preparer":{"model":"openai/gpt-6-luna","runId":"dcbadf68-223b-f4b7-387f-665d8213454f"},"reviewer":{"model":"anthropic/claude-opus-5.5","runId":"1799dcd3-b144-f789-379d-7c3626811715"}},"audit":{"status":"passed","reviewedAt":"2026-11-12T15:23:26.416Z","model":"anthropic/claude-opus-5.5"}},"canonicalUrl":"https://escapemeridian.com/readings/mit-technology-review-argues-gpt-5-s-health-push-raises-5adcf12e","formats":{"html":"/readings/mit-technology-review-argues-gpt-5-s-health-push-raises-5adcf12e","markdown":"/readings/mit-technology-review-argues-gpt-5-s-health-push-raises-5adcf12e.md","json":"/readings/mit-technology-review-argues-gpt-5-s-health-push-raises-5adcf12e.json"},"attribution":"Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.","maintainedGuideLinks":[],"readerContext":{"version":"reader-context-v1","attribution":"Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.","terms":[],"entities":[{"id":"openai","name":"OpenAI","kind":"organization","context":"OpenAI develops AI models and products and publishes research, release announcements and evaluations.","referenceUrls":["https://openai.com/news/"],"path":"/entities/openai","json":"/entities/openai.json","markdown":"/entities/openai.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"5adcf12ea176ab2a92d12d7eb1bae5e4be5e5ef9eb0bae09be3b55ec447c8b04","revision":"652f3ab1-99f1-cdae-fdbb-329f3a407f68","revisionKind":"publication-id","title":"MIT Technology Review argues GPT-5’s health push raises accountability questions","url":"https://escapemeridian.com/readings/mit-technology-review-argues-gpt-5-s-health-push-raises-5adcf12e"},"source":{"type":"source","id":"https://www.technologyreview.com/2025/08/12/1121565/what-you-may-have-missed-about-gpt-5/","revision":"819aa0cea6973dcfa266d265cb736d06490a7dd381a805f26cda541719faf251","revisionKind":"sha256-source-text","title":"What you may have missed about GPT-5","url":"https://www.technologyreview.com/2025/08/12/1121565/what-you-may-have-missed-about-gpt-5/"},"status":"available","items":[],"next":null,"scope":"Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter."},"revisions":[{"id":"5adcf12ea176ab2a92d12d7eb1bae5e4be5e5ef9eb0bae09be3b55ec447c8b04","title":"MIT Technology Review argues GPT-5’s health push raises accountability questions","argument":"MIT Technology Review published this article on August 12, 2025. It says GPT-5 has mostly underwhelmed: the release is more a product update than a breakthrough, with users highlighting mistakes and problems with the model’s automatic choice of which system to use. The article also notes that GPT-5 seems less likely to flatter users excessively.\n\nIts larger concern is OpenAI’s shift from promoting broad AI capabilities to encouraging particular uses, especially health advice. The publication describes OpenAI presenting a couple’s account of using ChatGPT to understand cancer diagnoses and consider treatment decisions. It argues that evidence about doctors using AI as a clinical tool does not establish that people without medical backgrounds should seek personal health advice from AI. It also points to a medical-journal report about a man who nearly died after a conversation with ChatGPT.\n\nThe article frames the unresolved stake as accountability. It says there is little indication that technology companies will be liable for harm caused by AI health advice, and leaves open what recourse users would have when harmful advice is given.","topicIds":["artificial-general-intelligence"],"domain":"other","claimType":"attributed reading","limits":"Text as retrieved on November 1, 2026; originally published August 12, 2025. The article does not establish how legal responsibility for harmful AI health advice would be determined. The quoted question about recourse is raised by Damien Williams.","speaker":{"name":"MIT Technology Review","roleAtSource":"Publisher"},"source":{"title":"What you may have missed about GPT-5","publisher":"MIT Technology Review","format":"Article","url":"https://www.technologyreview.com/2025/08/12/1121565/what-you-may-have-missed-about-gpt-5/","publishedAt":"2025-08-12T07:52:23Z","dateBasis":"Date from the publisher’s own records, through its API."},"evidence":[{"text":"MIT Technology Review characterizes GPT-5’s reception as mostly underwhelming and the release as a product update rather than a breakthrough.","excerpt":"Overall, as my colleague Grace Huckins pointed out, the new release represents more of a product update","url":"https://www.technologyreview.com/2025/08/12/1121565/what-you-may-have-missed-about-gpt-5/","relationship":"Source-stated"},{"text":"The publication argues that AI companies are increasingly promoting specific applications, and suggests one possible reason is a lack of expected technical breakthroughs.","excerpt":"A possible explanation for this pivot is that tech giants simply have not made the breakthroughs they’ve expected.","url":"https://www.technologyreview.com/2025/08/12/1121565/what-you-may-have-missed-about-gpt-5/","relationship":"Source-stated"},{"text":"The article says OpenAI began explicitly encouraging people to use its models for health advice with GPT-5’s launch.","excerpt":"With the launch of GPT-5, OpenAI has begun explicitly telling people to use its models for health advice.","url":"https://www.technologyreview.com/2025/08/12/1121565/what-you-may-have-missed-about-gpt-5/","relationship":"Source-stated"},{"text":"The publication argues that evidence about doctors benefiting from AI as a clinical tool does not establish that people without medical backgrounds should use AI for personal health advice.","excerpt":"it’s using evidence that doctors can benefit from AI as a clinical tool, as in the Kenya study, to suggest that people without any medical background should ask the AI model for advice about their own health.","url":"https://www.technologyreview.com/2025/08/12/1121565/what-you-may-have-missed-about-gpt-5/","relationship":"Source-stated"},{"text":"The article reports that a medical journal published a paper about a man who nearly died after a conversation with ChatGPT.","excerpt":"the Annals of Internal Medicine published a paper about a man who stopped eating salt and began ingesting dangerous amounts of bromide following a conversation with ChatGPT.","url":"https://www.technologyreview.com/2025/08/12/1121565/what-you-may-have-missed-about-gpt-5/","relationship":"Source-stated"},{"text":"The publication frames accountability and users’ recourse for harmful AI health advice as unresolved.","excerpt":"When ChatGPT gives you harmful medical advice because it’s been trained on prejudicial data, or because ‘hallucinations’ are inherent in the operations of the system, what’s your recourse?","url":"https://www.technologyreview.com/2025/08/12/1121565/what-you-may-have-missed-about-gpt-5/","relationship":"Source-stated"}],"sourceMedia":{"status":"unchecked","sourceUrl":"https://www.technologyreview.com/2025/08/12/1121565/what-you-may-have-missed-about-gpt-5/","checkedAt":null,"items":[]},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"When chatbot health advice causes harm, what recourse will users have, and who would be held responsible?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"MIT Technology Review argues GPT-5’s health push raises accountability questions","summary":"MIT Technology Review published this article on August 12, 2025. It says GPT-5 has mostly underwhelmed: the release is more a product update than a breakthrough, with users highlighting mistakes and problems with the model’s automatic choice of which system to use. The article also notes that GPT-5 seems less likely to flatter users excessively.\n\nIts larger concern is OpenAI’s shift from promoting broad AI capabilities to encouraging particular uses, especially health advice. The publication describes OpenAI presenting a couple’s account of using ChatGPT to understand cancer diagnoses and consider treatment decisions. It argues that evidence about doctors using AI as a clinical tool does not establish that people without medical backgrounds should seek personal health advice from AI. It also points to a medical-journal report about a man who nearly died after a conversation with ChatGPT.\n\nThe article frames the unresolved stake as accountability. It says there is little indication that technology companies will be liable for harm caused by AI health advice, and leaves open what recourse users would have when harmful advice is given.","question":"When chatbot health advice causes harm, what recourse will users have, and who would be held responsible?","claims":[{"text":"MIT Technology Review characterizes GPT-5’s reception as mostly underwhelming and the release as a product update rather than a breakthrough.","excerpt":"Overall, as my colleague Grace Huckins pointed out, the new release represents more of a product update","support":"source-stated"},{"text":"The publication argues that AI companies are increasingly promoting specific applications, and suggests one possible reason is a lack of expected technical breakthroughs.","excerpt":"A possible explanation for this pivot is that tech giants simply have not made the breakthroughs they’ve expected.","support":"source-stated"},{"text":"The article says OpenAI began explicitly encouraging people to use its models for health advice with GPT-5’s launch.","excerpt":"With the launch of GPT-5, OpenAI has begun explicitly telling people to use its models for health advice.","support":"source-stated"},{"text":"The publication argues that evidence about doctors benefiting from AI as a clinical tool does not establish that people without medical backgrounds should use AI for personal health advice.","excerpt":"it’s using evidence that doctors can benefit from AI as a clinical tool, as in the Kenya study, to suggest that people without any medical background should ask the AI model for advice about their own health.","support":"source-stated"},{"text":"The article reports that a medical journal published a paper about a man who nearly died after a conversation with ChatGPT.","excerpt":"the Annals of Internal Medicine published a paper about a man who stopped eating salt and began ingesting dangerous amounts of bromide following a conversation with ChatGPT.","support":"source-stated"},{"text":"The publication frames accountability and users’ recourse for harmful AI health advice as unresolved.","excerpt":"When ChatGPT gives you harmful medical advice because it’s been trained on prejudicial data, or because ‘hallucinations’ are inherent in the operations of the system, what’s your recourse?","support":"source-stated"}],"limitations":["Text as retrieved on November 1, 2026; originally published August 12, 2025.","The article does not establish how legal responsibility for harmful AI health advice would be determined. The quoted question about recourse is raised by Damien Williams."],"mediaId":null,"correctionReason":null,"id":"5adcf12ea176ab2a92d12d7eb1bae5e4be5e5ef9eb0bae09be3b55ec447c8b04","publicationId":"652f3ab1-99f1-cdae-fdbb-329f3a407f68","version":1,"sourceUrl":"https://www.technologyreview.com/2025/08/12/1121565/what-you-may-have-missed-about-gpt-5/","sourceTitle":"What you may have missed about GPT-5","publisher":"MIT Technology Review","publishedAt":"2025-08-12T07:52:23Z","originalPublished":{"value":"2025-08-12T07:52:23Z","precision":"instant","basis":"backfill:api_published"},"firstObservedAt":"2026-11-01T04:33:26.624Z","createdAt":"2026-11-12T15:28:25.506Z","historical":true,"entityIds":[],"topicIds":["artificial-general-intelligence"],"media":[],"sourceMedia":{"status":"unchecked","sourceUrl":"https://www.technologyreview.com/2025/08/12/1121565/what-you-may-have-missed-about-gpt-5/","checkedAt":null,"items":[]},"editorialDecision":{"outcome":"accept","mediaId":null,"inputHash":"f20da6947ace24fbafd3b554df084eb8b77e2efffd9279aa8c0bb93b7f3ec93e","rubricVersion":"unrecorded","provider":"typesafe","model":"jev-1.13.0","reasonCodes":["bridge_admitted_by_model"],"confidence":0.8,"decidedAt":"2026-11-01T04:33:23.819Z"},"provenance":{"textHash":"819aa0cea6973dcfa266d265cb736d06490a7dd381a805f26cda541719faf251","observationId":"bd889d9f70269c06300f273e94f6a165a1f326254f7c57d1500fccde33e51708","retrievedAt":"2026-11-01T04:35:20.190Z","kind":"publisher-page","preparer":{"model":"openai/gpt-6-luna","runId":"dcbadf68-223b-f4b7-387f-665d8213454f"},"reviewer":{"model":"anthropic/claude-opus-5.5","runId":"1799dcd3-b144-f789-379d-7c3626811715"}},"audit":{"status":"passed","reviewedAt":"2026-11-12T15:23:26.416Z","model":"anthropic/claude-opus-5.5"}}}],"audits":[{"publicationId":"652f3ab1-99f1-cdae-fdbb-329f3a407f68","verdict":"passed","findings":[{"dimension":"attribution","status":"pass","explanation":"The reading consistently attributes the article's assessments to MIT Technology Review, the publisher of this single newsletter-derived piece. The 'product update' view is credited to the publication, which fits because the author cites a Technology Review colleague. The limitations correctly note that the recourse question is Damien Williams's.","excerpt":"Overall, as my colleague Grace Huckins pointed out, the new release represents more of a product update"},{"dimension":"summary","status":"pass","explanation":"The reading leads with the article's verdict that GPT-5 mostly underwhelmed. It then explains the shift toward specific uses such as health advice, with the cancer-diagnosis example and the bromide case, and ends on accountability. It also keeps the positive note about less flattery. Terms are given in plain language, such as the automatic choice of which system to use. The title and question follow from the source.","excerpt":"So what’s the point of all this? Essentially, it’s about accountability."},{"dimension":"date","status":"pass","explanation":"August 12, 2025 matches the UTC day of originalPublished, and the reading states it at day precision. The sidebar items, such as the AlphaGo ten-year piece, appear to be later page additions, and the reading neither uses nor relies on them. There is no hindsight framing. modifiedAt is one second after publication, so the date is unaffected.","excerpt":"This story originally appeared in The Algorithm, our weekly newsletter on AI."},{"dimension":"support","status":"pass","explanation":"Each claim matches its excerpt and stays attributed. The bromide case is reported as a journal paper, followed by near death, as the source says. The pivot explanation keeps the source's 'possible' hedge. The question about recourse follows directly from the Williams quote and the article's statement on liability.","excerpt":"As things stand, there’s little indication tech companies will be made liable for the harm caused."},{"dimension":"media","status":"pass","explanation":"No media was selected and no verified candidates were supplied, so there is no media identity or relevance issue.","excerpt":null}],"createdAt":"2026-11-12T15:23:26.416Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}