Chronically implanted brain–computer interfaces can provide people with severe paralysis with reliable, independently usable communication over sustained everyday use.
Verification position derived from the record’s assessments; dates show when Faultline first recorded each stage.
Causal mechanisms recorded for this claim. The State Warrant above remains the authoritative current assessment.
Cross-participant clinical generalisation. Current high-performance systems are trained around individual neural anatomy, signal characteristics, attempted-speech patterns and clinical condition. Demonstrating excellent communication in one participant does not establish that comparable performance, calibration burden and functional benefit can be reproduced across people with different causes and degrees of paralysis. This is the load-bearing bottleneck between individual feasibility and a clinical communication modality.
Chronic interface stability and safety. Intracranial electrodes, tissue responses, connector integrity and recorded neural populations can change over time. A system may produce high short-term accuracy yet fail to preserve stable signals or impose an unacceptable surgical, infection or maintenance burden during long-term use.
Context dependence and neural non-stationarity. Prompted training tasks do not fully represent spontaneous conversation, concurrent behaviours, fatigue, posture, disease progression or day-to-day signal change. The 2026 multimodal study shows directly that decoders trained on isolated behaviours did not automatically generalise to simultaneous attempts, requiring context-inclusive data and continuing adaptation.
Operational dependency. Percutaneous connections, external computing hardware, specialist configuration, caregiver setup and recurrent supervised recalibration can prevent a technically capable decoder from becoming independently usable. Laboratory accuracy supports the governing claim only to the extent that the surrounding system can be operated reliably in everyday settings.
Replicated sustained independent communication. Implanted systems provide clinically meaningful spontaneous communication over months or years to multiple people with severe paralysis across independent clinical centres, with stable performance, manageable calibration and caregiver burden, acceptable implant safety and routine operation without continuous researcher intervention. Consistency across different causes of paralysis and materially different implant architectures would provide stronger resolution evidence.
Historical narrative recorded for this claim. It does not override the current State Warrant.
Questions retained in this record. The current State Warrant may have narrowed or reframed earlier questions.
Can sustained independent communication be reproduced across multiple participants, clinical centres and causes of paralysis rather than remaining dependent on exceptional single-participant results?
Raised 2026-09-15What combination of spontaneous communication accuracy, rate, correction burden and user-reported intelligibility should count as clinically meaningful everyday communication?
Raised 2026-09-15How stable are neural signals and functional performance over years as tissue response, disease progression, hardware ageing and daily context change?
Raised 2026-09-15How much caregiver setup, specialist maintenance and recalibration is compatible with the claim's requirement for independently usable communication?
Raised 2026-09-15Can context-inclusive and cross-participant training reduce calibration burden without allowing language-model priors to substitute plausible output for the user's intended communication?
Raised 2026-09-15| Mutation | Date | Field | Prior value | Current value |
|---|---|---|---|---|
| M-006 | 2026-09-28 | provenance_review_completed | — | LPR-001-D31 |
| M-005 | 2026-09-15 | diagnosis_held | — | DIAGNOSIS-HELD |
| M-004 | 2026-09-15 | mechanisms_recorded | — | MECHANISMS-RECORDED |
| M-003 | 2026-09-15 | assessment_issued | — | ASSESSMENT-ISSUED |
| M-002 | 2026-09-15 | instances_logged | — | INSTANCES-LOGGED |
| M-001 | 2026-09-15 | record_created | — | RECORD-CREATED |