The Anti-Backdoor (Anti-BAD) Challenge leaderboards display participants’ final results across all tracks and tasks. Scores are percentages (geometric mean of utility and backdoor mitigation), where higher is better. For evaluation metrics and scoring details, see the Challenge page.


Test Phase Results

Results announced on February 18, 2026.

Congratulations to all participants, and special thanks to everyone who contributed to making this challenge a success.

Rank Team Overall Gen Task 1 Gen Task 2 Cls Task 1 Cls Task 2 Mul Task 1 Mul Task 2
🥇AIST_ICT90.9890.8692.4395.4995.8395.0876.21
🥈UNIST87.0791.2992.7987.2687.6495.1168.33
🥉ptrcklv86.7689.6993.6483.9091.6994.3067.32
4ibaharul83.8589.7192.9279.4976.1093.5971.29
5trkosire83.6790.5993.4277.9275.0293.7571.29
6chrisbot7782.1891.1392.8076.1475.2094.9562.87
7nidala81.0091.2093.0876.3475.2089.8160.40
8DCL63.6491.2189.1176.5454.6222.5447.85
9summu9.060.000.000.000.000.0054.36
Clean reference87.4587.2989.4488.1591.8696.8271.14
Baseline (WAG)81.1591.2092.8775.7375.2089.4762.44

Notably, the top-ranked submission achieved an overall score surpassing the clean reference.

Two reference rows are included for context:


Development Phase Results

Final standings at the close of the development phase (November 7, 2025 – January 31, 2026).

Rank Team Overall Gen Task 1 Gen Task 2 Cls Task 1 Cls Task 2 Mul Task 1 Mul Task 2
1AIST_ICT93.1089.2690.7694.5097.4994.1192.49
2UNIST (seungb)91.5486.0791.8490.6194.1694.3992.16
3ptrcklv89.9784.1587.6791.2992.9992.5591.15
4DCL (chrisbot77)87.7082.8589.0589.5190.6984.3289.81
5nidala87.4782.0289.1589.5188.8085.6889.66
6DCL (abdellahelmrini)87.1682.6188.2388.9891.4382.2489.47
729986.9783.0188.9686.1789.1984.8589.64
8trkosire79.3581.9887.3189.3389.9141.0786.50
9ibaharul79.0982.2288.1588.6888.0840.8986.50
10gurubhandari77.4071.7073.2488.1577.9576.4076.95
11UNIST (kiwan_kwon)63.7484.6489.0373.1862.0628.7144.84
12Durinn AS54.6972.0571.1816.7350.3845.3572.43
13Robin12.410.000.0074.440.000.000.00
14qafig12.160.000.0072.980.000.000.00
Clean reference90.5184.4986.3390.3995.6694.8791.32
Baseline (WAG)87.3883.0388.6789.8688.2684.8589.64