In this regard, the presented paper asks a simple but urgent question about #LocationPrivacy under realistic assumptions.
Have you ever wondered whether #DifferentialPrivacy (DP) truly protects your location data once an attacker knows a bit more about the world around you?
© Dénes Erdős | ISTA
© Dénes Erdős | ISTA
Is privacy in the digital age possible?
Cynthia Dwork from Harvard University addressed this topic at the ISTA Lecture earlier this week. Her talk on #DifferentialPrivacy demonstrated how it safeguards personal data in industry and government.
Thanks to everyone who attended this event.
ELSA Board member @ahonkela.bsky.social contributed to the paper"Impact of Dataset Properties on Membership Inference Vulnerability of Deep Transfer Learning",presented at @neuripsconf.bsky.social'25
Article ➡️ www.helsinki.fi/en/faculty-s...
#MachineLearning #DifferentialPrivacy #PrivacyGuarantees
Launching today with #Oblivious an EU #CTF (Capture the Flag) competion on #Privacy-preserving #Database #Systems focused on #redteaming #DifferentialPrivacy.
Prizes total €7000 + Honorable mentions. Competition starts December 15th.
More info: developers.tiktok.com/blog/oblivious
OpenDP at IQSS has announced the launch of the Differential Privacy Deployments Registry, which will empower users to share details of their #differentialprivacy deployments and to develop best practices. The accompanying white paper is open for public comment until Dec 5.
opendp.org/2025/11/25/l...
Differential Privacy 101 in human words:
1️⃣ Take a batch of sequences
2️⃣ Clip each gradient so no single example can dominate
3️⃣ Add calibrated Gaussian noise 💉
4️⃣ Update weights
5️⃣ Repeat across a trillion tokens
#AI #Privacy #MachineLearning #DeepLearning #DifferentialPrivacy
📬 US-Regierung will Anonymität der Volkszählung aufheben
#Datenschutz #Netzpolitik #CensusBureau #COUNTAct #Deanonymisierung #differentialprivacy #HowardLutnick #Staatsbürgerschaft #TopDown
Republicans want to ban differential privacy in the Census, citing data distortion. But removing it could expose personal info, risking privacy for millions. It’s a battle over how we protect data in the age of AI.
www.wired.com/story/...
#DataPrivacy #Census #Policy #Ethics #DifferentialPrivacy
🔏𝗥𝗲𝘃𝗶𝗲𝘄 𝗗𝗮𝘆 𝗧𝘂𝗲𝘀𝗱𝗮𝘆!📖
#DifferentialPrivacy is an increasingly popular, though controversial, approach. This week, @benrothke.bsky.social provides a Hall of Fame recommendation of @simson.net 's 𝘋𝘪𝘧𝘧𝘦𝘳𝘦𝘯𝘵𝘪𝘢𝘭 𝘗𝘳𝘪𝘷𝘢𝘤𝘺, from @mitpress.bsky.social.
👉 tinyurl.com/2s3jk4cn
#CybersecurityBooks #DP
Concurrent Composition Theorems Boost Adaptive Differential Privacy
Concurrent composition theorems can improve adaptive differential privacy, researchers report. Read more: getnews.me/concurrent-composition-t... #differentialprivacy #adaptive
Differentially Private Algorithms for Wasserstein Barycenters
Researchers unveil the first differentially private algorithms for Wasserstein barycenters, delivering results on synthetic data, MNIST, and a U.S. population set. Read more: getnews.me/differentially-private-a... #differentialprivacy #wasserstein
The Great Screenshot Scandal: Microsoft Recall and Signal’s DRM Shield
dailysecurityreview.com/podcasts/the...
#ML #federatedlearning #differentialprivacy #edgeAI #QML #GDPR #privacy
Federated Learning www.geeksforgeeks.org/machine-lear...
Differential Privacy aimultiple.com/differential...
Gradient Denoising Improves Efficiency of Private LLM Fine‑Tuning
A denoising step after each DP‑SGD update restores low‑rank gradient structure, boosting sample efficiency when fine‑tuning RoBERTa on GLUE without compromising privacy. Read more: getnews.me/gradient-denoising-impro... #differentialprivacy #llm
Differential Privacy Scheme for Black‑Box Function Estimation
A new scheme enables private estimation of black-box functions without sensitivity bounds, letting users balance data against query count via a parameter; privacy loss stays within ε‑δ. getnews.me/differential-privacy-sch... #differentialprivacy #blackbox
Privacy Attacks and Countermeasures in Data‑Driven Systems
The paper classifies privacy attacks into three groups—anonymous re-identification, statistical reconstruction, and model inference—and notes defenses like noise perturbation. Read more: getnews.me/privacy-attacks-and-coun... #privacy #differentialprivacy
Backdoor Threat to Differential Privacy via Gaussian Pancake Mechanism
The Gaussian Pancake Mechanism (GPM) mimics the Gaussian mechanism but weakens differential‑privacy guarantees. 28 Sept 2025. Read more: getnews.me/backdoor-threat-to-diffe... #differentialprivacy #gaussianpancake #security
New Differentially Private Bootstrap Improves Statistical Inference
A new differentially‑private bootstrap provides valid confidence intervals, achieving nominal coverage on 2016 Canada Census data (reported 24 Sep 2025). getnews.me/new-differentially-priva... #differentialprivacy #bootstrap #statistics
Urania framework enables private AI chatbot analysis
Urania is a differential‑privacy framework for chatbot logs that uses private clustering, keyword extraction and histogram summarization, beating a non‑private Clio‑based baseline. Read more: getnews.me/urania-framework-enables... #differentialprivacy #ai #chatbot
Differentially Private Compression: New Sensitivity Bounds for LZ77
A new study shows LZ77’s global sensitivity is bounded by O(W^{2/3}·log n) and, when the window equals the input length, by Ω(n^{2/3}·log^{1/3} n). Presented at TCC 2025. Read more: getnews.me/differentially-private-c... #lz77 #differentialprivacy
2020 U.S. Census Privacy Revealed: Stronger Protection Than Expected
New research shows the 2020 U.S. Census added more noise than needed, estimating that noise levels could be reduced by 15.08%‑24.82% while preserving privacy. Read more: getnews.me/2020-u-s-census-privacy-... #census #differentialprivacy #privacy
New Monitoring Method Tracks Differential Privacy Violations Over Time
Monitoring aggregates deployment history to audit differential‑privacy guarantees, cutting required samples. Tests on Laplace and Gaussian mechanisms matched full‑audit accuracy. getnews.me/new-monitoring-method-tr... #differentialprivacy
Chernoff Differential Privacy Boosts Adversarial Classification
A new study introduces Chernoff differential privacy tying the privacy budget ε to tighter error exponents for classifiers, with the paper submitted on 15 March 2024. Read more: getnews.me/chernoff-differential-pr... #chernoffdp #differentialprivacy
Query-Efficient Private Hypothesis Selection via Scheffé Graph
A new LDP algorithm cuts hypothesis‑selection queries to ~O(k^{3/2}) non‑adaptive queries for k candidate distributions, improving on prior quadratic methods. Submitted September 2025. getnews.me/query-efficient-private-... #differentialprivacy #algorithm
LLMs Can Reconstruct Differentially Private Text, Study Shows
LLMs can reverse‑engineer DP‑sanitized text, achieving 91.20% recovery on ChatGPT‑4o and 94.01% on Claude‑3.5 in black‑box attacks. Read more: getnews.me/llms-can-reconstruct-dif... #largelanguagemodels #differentialprivacy #chatgpt4o
SynBench Benchmark Shows Gaps in Private Text Generation
SynBench, a differential‑privacy text generation benchmark, tests datasets and finds quality drops sharply on complex data, even for large LLMs. It adds an attack to reveal privacy leaks. getnews.me/synbench-benchmark-shows... #differentialprivacy #llms
Here's my latest blog alert on sublinear #algorithms and #differentialprivacy in a wide-ranging conversation with @ccanonne.github.io!
scieye.wordpress.com/2025/09/17/e...
#randomness #BigData #SciComm @simonsinstitute.bsky.social @sydney.edu.au
DPCheatSheet Helps Novice Developers Implement Differential Privacy with LLMs
DPCheatSheet is a web tool that helps novices add differential privacy to LLM‑generated code. In a study of six novices and three DP experts, users flagged more errors. getnews.me/dpcheatsheet-helps-novic... #differentialprivacy #llm
Google unveils VaultGemma, a high-performance private AI model
Google released VaultGemma, a 1‑billion‑parameter private LLM that matches non‑private models on benchmarks like MMLU. Code is on Hugging Face. Read more: getnews.me/google-unveils-vaultgemm... #vaultgemma #differentialprivacy #google
FedRP: Efficient Private Federated Learning via Random Projection
FedRP combines random projection with ADMM to give ε,δ differential‑privacy and drastically lower communication traffic. The paper was submitted on 12 Sep 2025. Read more: getnews.me/fedrp-efficient-private-... #federatedlearning #differentialprivacy
Another deep dive into #DifferentialPrivacy - #HIPAA #GDPR #PrivacyEngineering #PEPR open.substack.com/pub/masterso...
