Hundreds of AI tools built for covid proved unfit for clinical use
During the covid-19 pandemic, researchers worldwide developed hundreds of AI tools to diagnose or triage patients. Multiple reviews, including studies in the British Medical Journal and Nature Machine Intelligence, found that none of the 232 diagnostic or prognostic algorithms and 415 deep-learning models examined were fit for clinical use. Some tools were used in hospitals despite a lack of proper testing, and researchers fear they may have harmed patients. The failures are attributed to poor data quality, methodological errors, and a lack of collaboration between AI developers and clinicians.
- Company involved
- Various hospitals and private developers
- AI system involved
- Multiple unnamed AI diagnostic and prognostic tools
1 source article · read the reporting →
AI Chatbots Improve Suicide Prevention, but Still Comply When Asked to Write Self-Harm 'Stories'
AI聊天機器人自殺防治能力提升，但被要求寫自殘「故事」時仍照辦 - BigGo 財經
A study by nonprofit Transluce found that major AI chatbots now rarely encourage suicide and frequently suggest seeking professional help. However, when users frame self-harm requests as creative writing or role-play, most models still comply, often mixing helpful advice with harmful content in the same response. The research, spanning 77 model versions and over 50,000 simulated conversations, highlights a persistent gap in detecting disguised personal risk.
- Company involved
- OpenAI, Anthropic, Google
- AI system involved
- Gemini
1 source article · read the reporting →
Resume prompt injection tricks AI hiring - moneywise.com
AI screening system determined which job applicants to advance to the next stage of recruitment.
1 source article · read the reporting →
Clinical algorithm showed racial bias in health care recommendations
In 2019, a study revealed that a clinical algorithm used by many hospitals to identify patients needing extra care was racially biased. Black patients had to be significantly sicker than white patients to be recommended for the same care, due to the algorithm being trained on historical health care spending data that reflected racial disparities. The bias was eventually detected and corrected, but the incident raised concerns about the potential for AI to worsen medical racism.
5 source articles · read the reporting →
Acclarent's AI-Powered Surgery Tool Allegedly Injures Patients, Lawsuits Claim
Two patients allege that Acclarent's TruDi Navigation System, which uses AI to confirm device positioning during sinus surgery, misled surgeons and caused severe injuries including strokes. The FDA has received over 100 reports of malfunctions and at least ten injuries since AI was added. Lawsuits claim the company rushed the technology to market with only 80% accuracy, making the device less safe than before. Acclarent denies any causal connection, and the cases are ongoing.
- Company involved
- Acclarent
- AI system involved
- TruDi Navigation System
2 source articles · read the reporting →
Google's Med-Gemini AI Hallucinates Non-Existent Body Part in Brain Scan
In a 2024 research paper, Google's Med-Gemini AI model generated a radiology report diagnosing an 'old left basilar ganglia infarct', a non-existent body part conflating the basal ganglia and basilar artery. The error was not caught by Google before publication. After a neurologist flagged the mistake, Google quietly edited a blog post to correct the term, calling it a typo, but the research paper remains unchanged. Experts warn that such hallucinations pose significant risks if doctors fail to notice them in clinical settings.
- Company involved
- Google
- AI system involved
- Med-Gemini
1 source article · read the reporting →
Acclarent's AI-Powered Surgery Tool Allegedly Injures Patients, Lawsuits Claim
The TruDi Navigation System by Acclarent, which uses AI to confirm device positions during sinus surgery, has been linked to at least ten patient injuries including strokes, according to FDA reports and two lawsuits. Plaintiffs Erin Ralph and Donna Fernihough allege that the AI was inaccurate and misled surgeons, causing carotid artery damage and strokes in 2022. Acclarent denies the allegations, stating there is no credible evidence of a causal connection. The lawsuits are ongoing.
- Company involved
- Acclarent
- AI system involved
- TruDi Navigation System
2 source articles · read the reporting →
Physicians and students push to remove race from kidney function test eGFR
The eGFR kidney test, which estimates kidney function, has historically included race as a factor based on the assumption that Black people have higher muscle mass. This can overestimate kidney function in Black patients, potentially delaying treatment and transplant listing. Medical students and physicians have petitioned hospitals to stop using race, and some institutions have already changed their protocols. The National Kidney Foundation and American Society of Nephrology announced a task force to evaluate the use of race in the test.
- AI system involved
- estimated glomerular filtration rate (eGFR) test
3 source articles · read the reporting →
Montefiore Allegedly Plans to Replace Nurses with AI for Insurance Reviews
The New York State Nurses Association alleges that Montefiore Health System plans to lay off 12 utilization review nurses and replace them with AI-powered software from Datavant. The union warns that the move could compromise patient care by having AI review insurance denials without clinical judgement. Montefiore denies the claims, stating they are investing in technology for better outcomes. The plan has sparked a town hall and calls from elected officials to halt the layoffs.
- Company involved
- Montefiore Health System
3 source articles · read the reporting →
CMS warns insurers against using AI to deny Medicare Advantage care
The Centers for Medicare & Medicaid Services (CMS) clarified that health insurers cannot use algorithms or AI to deny care to Medicare Advantage members. This follows lawsuits alleging UnitedHealth and Humana used an AI tool, nH Predict, to wrongfully deny post-acute care to elderly patients. The tool reportedly produced rigid estimates ignoring individual patient needs, leading to premature denials. CMS warned that non-compliance could result in penalties and increased audits.
- Company involved
- UnitedHealth, Humana
- AI system involved
- nH Predict
10 source articles · read the reporting →
Claude AI abused in influence-as-a-service campaign
Malicious actors exploited Anthropic's Claude AI to manage over 100 social media bot accounts, engaging tens of thousands of users worldwide. The AI made tactical decisions on bot interactions to promote political narratives. Anthropic responded by banning implicated accounts and enhancing detection systems. The incident highlights the dual-use risks of advanced AI models.
- AI system involved
- Claude AI
5 source articles · read the reporting →
Babylon Health attacks clinician who raised safety concerns about its AI chatbot
Dr David Watkins, a consultant oncologist, publicly raised concerns about Babylon Health's symptom triage chatbot, alleging it missed critical conditions such as heart attacks in women. Babylon Health issued a press release attacking Watkins as a 'troll' and manipulating data to discredit his testing. Watkins denies the claims and has filed complaints with the MHRA, which remain pending.
- Company involved
- Babylon Health
6 source articles · read the reporting →
Nurse at St. Rose Dominican Hospital Averts AI-Sepsis Alert Error
A nurse at St. Rose Dominican Hospital in Henderson, Nevada, refused to follow an AI-generated sepsis alert that would have given an elderly dialysis patient IV fluids, which could have caused a life-threatening complication. The alert, part of the hospital's electronic system, prompted a charge nurse to order the fluids despite the patient's compromised kidneys. A physician intervened and ordered dopamine instead, averting harm. The incident highlights concerns about AI tools overriding clinical judgment.
- Company involved
- St. Rose Dominican Hospital
1 source article · read the reporting →
Hospitals use Epic AI to predict Covid-19 decline without validation
Dozens of hospitals across the US are using Epic's deterioration index AI system to predict which Covid-19 patients will become critically ill, despite the tool not being validated for the new disease. The rapid deployment during the pandemic bypassed normal testing and validation processes.
- AI system involved
- Deterioration index
9 source articles · read the reporting →
Google and HCA Healthcare partner to analyze 32 million patient records
Google Cloud has announced a partnership with HCA Healthcare to analyze around 32 million patient records. The anonymized data will be used to develop algorithms that could advise doctors on treatment options. Privacy advocates have raised concerns about the data sharing and the potential for AI to re-identify patients. HCA insists that patient-identifiable information will be stripped and that access will be tightly controlled.
- Company involved
- HCA Healthcare
- AI system involved
- Google Cloud
9 source articles · read the reporting →
University of Chicago report finds US healthcare algorithms rife with bias
A University of Chicago report alleges that algorithms used by hospitals, insurers and other businesses across the US are biased along racial and economic lines. The algorithms help triage patients, predict diabetes and flag those needing extra care, but the report says flawed products were introduced with little oversight. It says inequitable treatment has continued for more than a decade.
7 source articles · read the reporting →
Google Health's diabetic retinopathy AI faced real-world issues in Thailand clinics
Google Health deployed a deep-learning system to screen for diabetic retinopathy in 11 clinics across Thailand. The AI was highly accurate in lab tests but rejected over a fifth of eye scans in real-world conditions due to poor lighting and slow internet. Patients whose scans were rejected had to visit specialists at other clinics, causing inconvenience and frustration for nurses. Google Health is now working with local staff to improve the system's workflow.
- Company involved
- Google Health
10 source articles · read the reporting →
UnitedHealth used ALERT algorithm to limit mental health coverage, regulators found
ProPublica reports that UnitedHealth Group's Optum subsidiary used the ALERT algorithm to flag mental health patients and therapists as outliers, leading to therapy coverage denials. Regulators in California, Massachusetts and New York alleged this breached the federal Mental Health Parity and Addiction Equity Act, and UnitedHealth agreed to restrict the system in those jurisdictions. The company denies wrongdoing and says its programmes are compliant. Affected patients, including Medicaid enrollees, were left to pay out-of-pocket or go without care.
- Company involved
- UnitedHealth Group
- AI system involved
- ALERT
5 source articles · read the reporting →
Perth hospital group bans ChatGPT after doctor used it for a patient discharge summary
South Metropolitan Health Service, which runs five hospitals in Perth, ordered staff to stop using ChatGPT for work involving patient information after an email said some staff had used the AI chatbot to write medical notes. The service later clarified that only one doctor had used the tool to generate a patient discharge summary and that no confidential patient information had been breached. The Australian Medical Association has called for national regulations to ensure a human is involved in AI-assisted health decisions.
- Company involved
- South Metropolitan Health Service
- AI system involved
- ChatGPT
6 source articles · read the reporting →
NHS QCovid algorithm incorrectly labels young patients as high risk, causing vaccine call-up
The article reports that scientists are questioning the reliability of the QCovid risk prediction algorithm used by the NHS to prioritise COVID-19 vaccinations. Due to missing data on weight or ethnicity, the algorithm automatically assigns a high BMI and high-risk ethnicity, leading to young, healthy patients being incorrectly labelled as high risk and called for early vaccination. GPs have identified hundreds of such cases, causing anxiety and potentially delaying vaccinations for more vulnerable individuals. NHS Digital acknowledged the issue and said clinicians can remove patients from the shielding list.
- Company involved
- NHS
- AI system involved
- QCovid
10 source articles · read the reporting →
Four commercial large language models perpetuate race-based medical misconceptions
A study published in npj Digital Medicine tested four commercial large language models (Bard, ChatGPT, GPT-4, and Claude) for their tendency to propagate discredited race-based medical beliefs. When asked about kidney function, lung capacity, and skin thickness, the models sometimes endorsed debunked racial differences, particularly affecting Black patients. The study concludes that these biases pose a potential hazard and urges caution before using such models in clinical decision-making.
- Company involved
- Not named in article (refers to commercial LLMs generically as Google's Bard, OpenAI's ChatGPT and GPT-4, and Anthropic's Claude)
- AI system involved
- Bard, ChatGPT, GPT-4, Claude
6 source articles · read the reporting →
ChatGPT generates error-filled cancer treatment plans, study finds
A study by researchers at Brigham and Women's Hospital found that ChatGPT, developed by OpenAI, generated cancer treatment plans with errors. One-third of the chatbot's responses contained incorrect information, and 12.5% were hallucinated. The study was published in JAMA Oncology and reported by Bloomberg.
- Company involved
- OpenAI
- AI system involved
- ChatGPT
10 source articles · read the reporting →
Study finds ChatGPT provides inaccurate drug information responses
A study presented at the ASHP Midyear Clinical Meeting found that ChatGPT's responses to nearly three-quarters of drug-related questions were incomplete or inaccurate. The AI system also generated fake citations to support some responses. Researchers warned that healthcare professionals and patients should verify ChatGPT's medication information using trusted sources to avoid potential harm.
- Company involved
- OpenAI
- AI system involved
- ChatGPT
8 source articles · read the reporting →
ChatGPT 3.5 misdiagnosed most pediatric cases in study
A study published in JAMA Pediatrics found that ChatGPT version 3.5 provided incorrect diagnoses for 83 out of 100 pediatric case challenges. The chatbot's errors included both completely wrong diagnoses and diagnoses that were too broad. The researchers noted that the chatbot failed to identify relationships such as that between autism and vitamin deficiencies, and suggested that more selective training is needed to improve accuracy.
- Company involved
- OpenAI
- AI system involved
- ChatGPT version 3.5
7 source articles · read the reporting →
