IBM Watson for Oncology provided unsafe cancer treatment recommendations
IBM Watson for Oncology provided clinically unsafe and incorrect treatment recommendations to healthcare providers. The system allegedly suggested dangerous treatments, such as bleeding drugs for patients with severe hemorrhage.
Watson frequently gave erroneous cancer treatment advice, such as prescribing bleeding drugs for a patient with severe bleeding.
Key facts
- What
- IBM Watson for Oncology provided clinically unsafe and incorrect treatment recommendations to healthcare providers.
- Incident date
- Jan 1, 2018
- Who
- IBM Watson Health
- Failure mode
- Hallucination
- AI surface
- Algorithmic Decision
- Severity
- High
What happened
Internal IBM documents and reports from STAT News revealed that Watson for Oncology frequently gave erroneous cancer treatment advice. In certain instances, the system recommended drugs that could cause bleeding for patients already experiencing severe bleeding. These failures occurred during its deployment in medical environments, including a partnership with the University of Texas M.D.
What broke inside the model
- 01 · TriggerA user asks for a fact, a citation, or a figure.
- 02 · Model stepThe model writes a fluent, confident answer.
- 03 · Control gapNothing ties the claim back to a real source.
- 04 · FailureA fabricated fact ships as if it were verified.
- 05 · ConsequenceThe false claim reaches a customer, a court, or the public.
Confidence holds, and even spikes, as the claim detaches from any source.
The system failed to correctly interpret clinical data in context, often relying on synthetic data cases rather than real patient evidence. This resulted in a failure to synthesize medical knowledge into safe, patient-specific treatment paths.
What it cost
Sources
- SocialIBM's Watson recommended unsafe and incorrect cancer treatmentsstatnews.com
- PressReport: IBM Watson delivered 'unsafe and inaccurate' cancer recommendationsmassdevice.com
Cite this entry
https://failureindex.ai/failures/ibm-watson-oncology-provided-unsafe-cancerAI Failure Index. "IBM Watson for Oncology provided unsafe cancer treatment recommendations" (FI-0359). Realm Labs. https://failureindex.ai/failures/ibm-watson-oncology-provided-unsafe-cancer (indexed Jun 9, 2026).Data fields CC-BY 4.0, prose citation permitted. Incident ID FI-0359. Full dataset at /data.
Note from Realm Labs, the Index steward
How Realm fits
- Prism
- OmniGuard
- AI Detection & Response (AIDR)
This entry sits in the index's predictive wing: a system that scores, ranks, perceives, or steers rather than generates. Realm's runtime layer is built for the generative and agentic systems now moving into these same decision seats, where it watches a model's internal state and holds an unsupported claim or an unchecked action before it commits. The control gap on this record, an automated decision that reached people with no runtime check in front of it, is the same gap. The index keeps predictive failures on the record because the pattern carries straight into the systems shipping today.