Data61 AI-security researcher: adversarial robustness, privacy-preserving retrieval, and agentic AI threat analysis
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
21
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-16) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
Not yet scored. The complete works list underlying an output axis (publication volume and productivity trend) is unavailable; the raw record lists 80 works over the last three years, but the output dimension itself has not been computed.
Impact—
Not yet scored. Citation-level influence data (citation counts and citing works) is insufficient here; the raw record carries an h-index of 21 and an i10-index of 42, but the impact dimension score has not been computed.
Focus0/100
The focus score is 0, the lowest on the 0-100 scale, reflecting wide topical dispersion rather than concentration. The field distribution splits evenly at 0.5/0.5 between privacy-preserving information retrieval and trustworthy-ML security, and the raw topic record spans five clusters: adversarial robustness (59), privacy-preserving technologies (36), digital media forensic detection (30), security and verification (26), and spam/phishing detection (19).
Collaboration48/100
Collaboration scores 48 on the 0-100 scale, a mid-range value. The raw record lists 10 distinct co-authors, and the recurring set includes Xinguo Feng, Zhongkui Ma, Zihan Wang, Chuan Yan, Guowei Yang, Guangdong Bai, Heewon Baek, Kristen Moore, Hyoungshick Kim, and Surya Nepal — a stable but moderately sized core network rather than a large rotating collaboration base.
Activity100/100
Activity reaches 100, the top of the scale. Works per year climb steadily from 11 (2020) through 13 (2021), 15 (2022), 19 (2023), and 24 (2024) to 29 (2025), with 27 already recorded in 2026; the raw record counts 80 works in the last three years, confirming a high and still-rising publication cadence.
Reproducibility55/100
Reproducibility scores 55 (the record's reproMean) across 2 scored papers. Both Evolve-reviewed 2026 arXiv papers carry a medium repro-readiness label: the agentic AI security survey is judged rigorous and honest but built on a small empirical base (22 artifact-backed papers of 66 studies), while SHAQ's privacy results are tested only on 32-token documents.
Research summary
Alsharif Abuadbba is a researcher at Data61 in Australia working across two evenly split fields — AI security / privacy-preserving information retrieval and AI security (computer security / trustworthy ML), each at 0.5 in the profile's field distribution. The OpenAlex-resolved record shows an h-index of 21 and an i10-index of 42, with 80 works in the last three years. Activity is the strongest axis (score 100): works per year climb from 11 in 2020 through 19 in 2023 to 29 in 2025, with 27 already recorded in 2026. Topic counts span adversarial robustness in machine learning (59), privacy-preserving technologies (36), digital media forensic detection (30), security and verification in computing (26), and spam and phishing detection (19); a focus score of 0 flags this breadth rather than concentration. Evolve has reviewed two 2026 arXiv papers: a cross-dimensional threat taxonomy for agentic AI security covering 66 studies with 22 artifact-backed papers, and SHAQ (Shadow Queries), which replaces document embeddings with shadow-query embeddings to cut vec2text recovery to about 0.27 R-1 at near-baseline utility, tested so far only on 32-token documents. Collaboration is mid-range (score 48, 10 distinct co-authors), and the output and impact dimension scores are not yet computed.
Collaboration pattern
Collaboration sits mid-range (score 48) with 10 distinct co-authors in the raw record. The recurring set spans Xinguo Feng, Zhongkui Ma, Zihan Wang, Chuan Yan, Guowei Yang, Guangdong Bai, Heewon Baek, Kristen Moore, Hyoungshick Kim, and Surya Nepal — a stable core rather than a wide rotating pool. Both Evolve-reviewed 2026 papers place the researcher in a middle author position without corresponding status, consistent with contribution within multi-author teams.
Research keywords
Highlights
Cross-dimensional threat taxonomy for agentic AI security
SHAQ: shadow queries blunt embedding inversion in vector databases
Evaluation maturity gap in agentic AI security research
Timeline
- 2019Earliest year in the OpenAlex counts-by-year record, with 1 work; an anomalous earlier entry for 2000 (3 works) is treated as a metadata artifact.
- 2020Output accelerates: 11 works recorded, beginning a steady multi-year climb in the annual counts.
- 202319 works recorded, the mid-point of the steady climb through the 2020s.
- 2025Peak of the series so far: 29 works recorded in a single year.
- 202627 works recorded year-to-date, including both Evolve-reviewed papers — the agentic AI security threat taxonomy survey and SHAQ shadow queries for private vector retrieval.
Papers in the Evolve catalog
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- Bibliometric dimensions are computed from OpenAlex-profiled data; the output and impact dimension scores are not yet computed, though h-index (21) and i10-index (42) appear in the raw record without a stated source.
- The counts-by-year series includes an anomalous 2000 entry (3 works) inconsistent with the continuous 2019-2026 run; trend readings here exclude it.
- The reproducibility score (reproMean 55) rests on only 2 scored papers, and Evolve's qualitative review covers just two 2026 arXiv preprints, neither peer-reviewed at review time.
- 2026 work counts are year-to-date at the snapshot and may change; this profile is name-resolved from public bibliographic data (OpenAlex, ORCID 0000-0001-9695-7947).