From 73d5f4cda77e98b1e13d2d9c131114bd88689b83 Mon Sep 17 00:00:00 2001 From: Phillis Tang Date: Wed, 22 Jan 2025 14:52:30 -0800 Subject: [PATCH 1/3] update privacy readme for opSupportLimits --- security-privacy.md | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/security-privacy.md b/security-privacy.md index cd270204..661d416a 100644 --- a/security-privacy.md +++ b/security-privacy.md @@ -26,7 +26,9 @@ No. > - 2.6 What information from the underlying platform, e.g. configuration data, is exposed by this specification to an origin? -No information from the underlying platform is exposed directly. An execution time analysis may reveal indirectly the performance of the underlying platform's neural network hardware acceleration capabilities relative to another underlying platform. +The [opSupportLimits()](https://www.w3.org/TR/webnn/#api-mlcontext-opsupportlimits) exposes op level feature support limits that are different across backends. The information exposed roughly correspond to the underlying operating system version so it shouldn't introduce more entropy. + +Additionally an execution time analysis may reveal indirectly the performance of the underlying platform's neural network hardware acceleration capabilities relative to another underlying platform. > - 2.7 Does this specification allow an origin access to sensors on a user’s device From 7bf74d5ec75a2ffba4b9ec36596068bf2ac45df5 Mon Sep 17 00:00:00 2001 From: Yajing Tang Date: Thu, 23 Jan 2025 09:28:28 -0800 Subject: [PATCH 2/3] Update security-privacy.md Co-authored-by: Dwayne Robinson --- security-privacy.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/security-privacy.md b/security-privacy.md index 661d416a..234e550b 100644 --- a/security-privacy.md +++ b/security-privacy.md @@ -26,7 +26,7 @@ No. > - 2.6 What information from the underlying platform, e.g. configuration data, is exposed by this specification to an origin? -The [opSupportLimits()](https://www.w3.org/TR/webnn/#api-mlcontext-opsupportlimits) exposes op level feature support limits that are different across backends. The information exposed roughly correspond to the underlying operating system version so it shouldn't introduce more entropy. +The [opSupportLimits()](https://www.w3.org/TR/webnn/#api-mlcontext-opsupportlimits) exposes per-operator feature support limits that are different across backends. The information exposed roughly corresponds to the underlying operating system version, and so it shouldn't introduce more entropy. Additionally an execution time analysis may reveal indirectly the performance of the underlying platform's neural network hardware acceleration capabilities relative to another underlying platform. From a32ffd59b59db1d654eabc9f2b309dec477dc760 Mon Sep 17 00:00:00 2001 From: Yajing Tang Date: Thu, 23 Jan 2025 11:10:57 -0800 Subject: [PATCH 3/3] Update security-privacy.md Co-authored-by: Joshua Bell --- security-privacy.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/security-privacy.md b/security-privacy.md index 234e550b..0afa6b27 100644 --- a/security-privacy.md +++ b/security-privacy.md @@ -26,9 +26,9 @@ No. > - 2.6 What information from the underlying platform, e.g. configuration data, is exposed by this specification to an origin? -The [opSupportLimits()](https://www.w3.org/TR/webnn/#api-mlcontext-opsupportlimits) exposes per-operator feature support limits that are different across backends. The information exposed roughly corresponds to the underlying operating system version, and so it shouldn't introduce more entropy. +The [opSupportLimits()](https://www.w3.org/TR/webnn/#api-mlcontext-opsupportlimits) method intentionally exposes per-operator behavior that is expected to vary depending on the underlying platform, such as the supported data types or preferred data layout for each operator. The information exposed roughly corresponds to the underlying operating system and/or browser version, and so it shouldn't introduce additional entropy. -Additionally an execution time analysis may reveal indirectly the performance of the underlying platform's neural network hardware acceleration capabilities relative to another underlying platform. +Additionally, an analysis of execution time may indirectly reveal the performance of the underlying platform's neural network hardware acceleration capabilities relative to another underlying platform. > - 2.7 Does this specification allow an origin access to sensors on a user’s device