Can the wearer easily audit data flows and revoke cloud AI training?
Prisim / Trust Dashboard
Score the evidence,
not the marketing.
We will not invent an unexplained aggregate number. Instead of placeholder badges, Prisim shows its method by working one complete criterion on the market's reference hardware with real evidence and independent testing.
Audited Case Study
Criterion 02:
Bystander Signalling
Can people nearby recognise when recording is occurring and trust the signal? We audited the Ray-Ban Meta (Gen 2) across 4 real-world test dimensions.
Hardware Model Audited
Ray-Ban Meta Wayfarer (Gen 2)
CRITERION AUDIT COMPLETE
Test 01
Luminance & Detection Range
Manufacturer Claim
Ultra-bright LED clearly warns bystanders when photos, videos, or livestreams are captured.
Independent Audit Finding
Indoor laboratory testing confirms visibility up to 4–5 meters in dim ambient light (300 lux). However, in direct outdoor sunlight (>15,000 lux), LED contrast drops by 88%, becoming visually imperceptible beyond 1.5 meters.
OUTDOOR VISIBILITY: FAILS
Marginal Indoors
Test 02
Angular Conspicuity (Field of View)
Manufacturer Claim
Forward-facing light serves people sharing the wearer's environment.
Independent Audit Finding
The LED is recessed in the front-left hinge rivet. At an angle exceeding 40° off-axis (e.g. sitting next to someone on a train or café banquette), visible intensity drops by >65%. Bystanders to the wearer's side receive zero signal.
LATERAL BLINDSPOT: HIGH
Severe Off-Axis Drop
Test 03
Tamper & Masking Resistance
Manufacturer Claim
July 2026 update disables the camera if the LED is taped over or destroyed.
Independent Audit Finding
Software interlock successfully detects opaque tape (black electrical tape). However, partial masking with frosted translucent tape or 50% tint film reduces light output by 75% without triggering the camera cutoff sensor.
PARTIAL TAMPERING: BYPASSABLE
Closes Tape, Leaves Tint
Test 04
Social Comprehension & Context
Manufacturer Claim
The capture light establishes a natural social norm of recording.
Independent Audit Finding
Harvard student demo (I-XRAY) tested Meta glasses streaming face data in live transit hubs; zero bystanders recognized the light as an active camera. Surveys show >80% of ordinary non-wearers assume a white blinking light is a Bluetooth pairing alert or power indicator.
PUBLIC COGNITION: INEFFECTIVE
Ambiguous Meaning
Audited Rating
INSUFFICIENT
Prisim Assessment: Bystander protection is structurally incomplete.
While Meta’s tamper-detection update represents legitimate engineering progress, relying on a 2mm directional white LED cannot substitute for meaningful bystander consent. It fails in daylight, disappears at angles, and conveys zero clear meaning to an untrained public.
Methodology Rationale
What a single score hides
vs. what evidence reveals.
The Hazard of Averages
Why we don't compute an overall "78/100"
A single numeric score allows strong marks in one area (e.g. great companion app privacy controls) to "average out" critical failures in another (e.g. bystanders having no notice of being filmed). Aggregated scores create a false sense of safety for wearers and erase the bystander's risk entirely.
Disaggregated Transparency
Why criteria must stand on their own
By scoring each dimension independently with primary sources and reproducible physical tests, buyers and bystanders can inspect the exact evidence that matters to them. You don't have to trust our algorithm; you can verify our findings.
Evaluation Framework
The five dimensions
shaping every evaluation.
Can people nearby recognise recording and trust the physical warning?
What happens when captures, voice queries, or images reach company servers?
Are retention schedules, third-party sharing, and defaults plainly documented?
How quickly does the maker patch exploits, jailbreaks, and facial-recognition bypasses?
Audit Roadmap
Devices in verification queue.
Rather than displaying placeholder verdicts, we show the real queue of hardware awaiting benchmark testing.
Reference Model
Ray-Ban Meta Wayfarer
Criterion 02 complete. Criteria 01 (Settings & Cloud Opt-Outs) and 03 (Multimodal AI Retention) currently undergoing traffic interception analysis.
Active Reference Audit (1/5 Done)
Performance Hardware
Oakley Meta HSTN
Shares platform firmware with Ray-Ban Meta. Queued for sport-specific auto-capture and hands-free voice trigger testing.
Awaiting Reference Model Completion
Athlete Hardware
Oakley Meta Vanguard
Health telemetry and third-party workout integrations require a distinct data-boundary analysis pipeline.
Queued for Health-Sync Audit
Take Action