google/differential-privacy

Issues somebody may already be working on are listed last. List everything by date

• Java's thresholded count underestimates δ because of the same continuous/discrete distribution mismatch as in Go

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 4/5 3-5 days Newbie friendliness 48/100

  google/differential-privacy#499 · Opened Sep 28, 2026

• Privacy on Beam doesn't track the privacy budget for many operations

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 4/5 3-5 days Newbie friendliness 48/100

  google/differential-privacy#498 · Opened Sep 28, 2026

• `dp_accounting` largely doesn't validate its floating-point inputs, which sometimes fails in potentially problematic ways

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 5/5 Over a week Newbie friendliness 35/100

  google/differential-privacy#497 · Opened Sep 28, 2026

• Under Poisson sampling, the `PLDAccountant` composes the inner event both before and after sampling

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 2/5 Half a day Newbie friendliness 78/100

  google/differential-privacy#496 · Opened Sep 28, 2026

• Java's Gaussian threshold δ can be wrong by a non-trivial factor due to floating-point imprecision

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 4/5 3-5 days Newbie friendliness 48/100

  google/differential-privacy#495 · Opened Sep 28, 2026

• The `RDPAccountant` with repeat-and-select mechanisms reports the δ of the inner mechanism regardless of how many times it's repeated

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 4/5 3-5 days Newbie friendliness 48/100

  google/differential-privacy#494 · Opened Sep 28, 2026

• The `RDPAccountant` ignores `count` when composing exponential mechanisms

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 3/5 1-2 days Newbie friendliness 62/100

  google/differential-privacy#493 · Opened Sep 28, 2026

• The `PLDAccountant` silently accepts unsupported Poisson-subsambled events and charges nothing for them

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 3/5 1-2 days Newbie friendliness 65/100

  google/differential-privacy#492 · Opened Sep 28, 2026

• The `PLDAccountant` ignores `count` when composing randomized response events

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 2/5 1-3 hours Newbie friendliness 88/100

  google/differential-privacy#491 · Opened Sep 28, 2026

• Go's `BoundedSumInt64` accepts both `MaxContributions` and `Upper`, but skips clamping when both are set (!)

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 3/5 1-2 days Newbie friendliness 72/100

  google/differential-privacy#490 · Opened Sep 28, 2026

• In Java's `LongBoundedSum`, setting `lower` to `Long.MIN_VALUE` under-estimates the sensitivity

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 2/5 1-3 hours Newbie friendliness 73/100

  google/differential-privacy#489 · Opened Sep 28, 2026

• Go's `BoundedSumInt64` used with `MaxContributions` incorrectly assumes that each user contributes to a single partition when computing thresholds

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 4/5 3-5 days Newbie friendliness 45/100

  google/differential-privacy#488 · Opened Sep 24, 2026

• Go's `Count.ThresholdedResult` computes the threshold from the continuous distribution but rounds to integers; leading to an underestimate in δ

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 4/5 3-5 days Newbie friendliness 45/100

  google/differential-privacy#487 · Opened Sep 24, 2026

• Go's `BoundedSumInt64` isn't safe against overflows in the l0×lInf multiplication

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 3/5 1-2 days Newbie friendliness 65/100

  google/differential-privacy#486 · Opened Sep 24, 2026

• On amd64, Go thresholding keeps every partition if the threshold overflows to +Inf

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 4/5 3-5 days Newbie friendliness 45/100

  google/differential-privacy#485 · Opened Sep 24, 2026

• C++ noise addition relies on a randomness buffer shared among forks, leading to identical noise across forks

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 4/5 3-5 days Newbie friendliness 40/100

  google/differential-privacy#484 · Opened Sep 24, 2026

• Go Count's `PreThresholdedResult` doesn't check if the result has already been released

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 3/5 1-2 days Newbie friendliness 65/100

  google/differential-privacy#483 · Opened Sep 24, 2026

• The README is inconsistent with the attack model regarding floating-point non-associativity

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 2/5 1-3 hours Newbie friendliness 70/100

  google/differential-privacy#482 · Opened Sep 23, 2026

• Data Sok

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 5/5 Over a week Newbie friendliness 10/100

  google/differential-privacy#467 · Opened Aug 13, 2026

• Help

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 5/5 Over a week Newbie friendliness 1/100

  google/differential-privacy#465 · Opened Aug 3, 2026

• API docs for dp-accounting

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 3/5 1-2 days Newbie friendliness 52/100

  google/differential-privacy#443 · 3 comments · Opened Jun 11, 2026

• [API Gap] Marginal scalar accounting yields optimistic estimates for jointly observed correlated Gaussian releases

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 2/5 1-3 hours Newbie friendliness 58/100

  google/differential-privacy#428 · 1 comment · Opened May 20, 2026

• [email protected]

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 5/5 Over a week Newbie friendliness 1/100

  google/differential-privacy#400 · 1 comment · 1 reaction · Opened Mar 22, 2026

• RFC: dp accounting should use rust as its backend

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 5/5 Over a week Newbie friendliness 25/100

  google/differential-privacy#397 · Opened Feb 15, 2026

• Add support for parallel composition in dp-accounting

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 5/5 Over a week Newbie friendliness 30/100

  google/differential-privacy#379 · 2 comments · Opened Jan 7, 2026

• Sensitivity off by a factor of 2 with 'replace_one' neighbouring relation

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 4/5 3-5 days Newbie friendliness 45/100

  google/differential-privacy#378 · 4 comments · Opened Dec 18, 2025

• Add support for aggregating multiple values

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  pipelinedp4j

  Difficulty 5/5 Over a week Newbie friendliness 30/100

  google/differential-privacy#285 · 1 reaction · Opened Feb 6, 2025

• Feature: Double Mixture of Gaussians for dp_accounting

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 5/5 Over a week Newbie friendliness 35/100

  google/differential-privacy#274 · 2 comments · Opened Nov 4, 2024

• Question on monotonic queries

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 5/5 Over a week Newbie friendliness 25/100

  google/differential-privacy#272 · 5 comments · Opened Sep 17, 2024

• Export TestNoiseFactory in Maven artifact

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 4/5 3-5 days Newbie friendliness 35/100

  google/differential-privacy#267 · Opened Apr 24, 2024

• Expose RNG or Seeded mechanisms outside of `-testing`

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 5/5 Over a week Newbie friendliness 20/100

  google/differential-privacy#107 · Opened Feb 8, 2022

• Add CMake support to the C++ library

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  enhancement

  Difficulty 4/5 3-5 days Newbie friendliness 35/100

  google/differential-privacy#62 · 1 comment · 2 reactions · Opened Nov 18, 2020

• C++ Proposal: Return doubles

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 5/5 Over a week Newbie friendliness 25/100

  google/differential-privacy#60 · Opened Oct 23, 2020

• Improving comments in privacy-critical code blocks.

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 2/5 1-3 hours Newbie friendliness 50/100

  google/differential-privacy#16 · Opened Jan 23, 2020

• Clearly distinguish private data and privitized data by using different variable naming conventions on each side of the noise barrier.

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 4/5 3-5 days Newbie friendliness 25/100

  google/differential-privacy#13 · 4 comments · Opened Jan 6, 2020

• Support for RDRAND

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 5/5 Over a week Newbie friendliness 20/100

  google/differential-privacy#12 · 3 comments · Opened Jan 6, 2020