Summary
Brave Leo has repeatedly produced factually inaccurate, misleading, contradictory, or insufficiently verified responses, experienced through both Brave Browser’s integrated Leo AI functionality and Leo AI on Windows Desktop.
This is NOT a single hallucination. The documented conversation includes incorrect information, references to nonexistent Brave plans, misleading Email Protection guidance, contradictory answers and repeated troubleshooting that resulted in several dead ends.
Observed Behavior
- Leo presents uncertain information as factual.
- Leo provides incorrect/outdated Brave-specific information.
- Leo references products/plans/options that cannot be verified.
- Leo contradicts earlier responses.
- Leo sent me through troubleshooting workflows that do not resolve the issue.
- Corrections frequently require multiple exchanges rather than being identified proactively.
The transcript specifically records the resulting pattern as wrong answers, misleading information, wasted time, contradictions and degraded trust.
Impact:
Significant wasted troubleshooting time for nothing.
Frequent user confusion.
Repeated and very frustrating dead ends.
Excessive corrective back-and-forth.
Serious reduction in trust in Leo’s factual reliability, accuracy and effectiveness.
Expected Behavior
Leo should:
- verify Brave-specific information against current authoritative sources,
- distinguish verified facts from uncertainty,
- never fabricate/mislabel plans, features, settings, or product behavior,
- detect contradictions with its prior responses,
- avoid directing users through invalid or nonexistent workflows.
Request:
Please investigate this as a Brave Leo factual-reliability and response-quality issue, rather than only as an Email Protection/UI problem, because the underlying concern is the repeated pattern of inaccurate and misleading AI guidance across multiple interactions and Leo access methods.
Thank you in advance.
