https://seppia978.github.io/research/2026-07-01T00:00:00+00:00weeklyhttps://seppia978.github.io/publication/2606-arxiv-finetuning_reversion/2026-06-26T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/ai-safety/2026-06-26T00:00:00+00:00weeklyhttps://seppia978.github.io/publication_types/article/2026-06-26T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/fine-tuning-reversion/2026-06-26T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/mechanistic-understanding/2026-06-26T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/model-adaptation/2026-06-26T00:00:00+00:00weeklyhttps://seppia978.github.io/publication_types/2026-06-26T00:00:00+00:00weeklyhttps://seppia978.github.io/publication/2026-06-26T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/2026-06-26T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/chain-of-thought/2026-06-09T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/multi-turn-reasoning/2026-06-09T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/safety-evaluation/2026-06-09T00:00:00+00:00weeklyhttps://seppia978.github.io/publication/2606-arxiv-cot_knows_better/2026-06-09T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/ai-safety-benchmarks/2026-06-01T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/nsfw/2026-06-01T00:00:00+00:00weeklyhttps://seppia978.github.io/publication_types/paper-conference/2026-06-01T00:00:00+00:00weeklyhttps://seppia978.github.io/publication/2511-eccv-spqr/2026-06-01T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/t2i-diffusion-models/2026-06-01T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/activation-watermarking/2026-03-24T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/adaptive-attacks/2026-03-24T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/llms-safety/2026-03-24T00:00:00+00:00weeklyhttps://seppia978.github.io/publication/2603-arxiv-activation_watermarking/2026-03-24T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/safety-monitoring/2026-03-24T00:00:00+00:00weeklyhttps://seppia978.github.io/event/mbzuai_sec_symposium_2602/2026-02-09T00:00:00+00:00weeklyhttps://seppia978.github.io/event/2026-02-09T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/spqr/2026-02-09T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/first-token-probability/2026-01-01T00:00:00+00:00weeklyhttps://seppia978.github.io/publication/2505-ftp_prefatt/2026-01-01T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/mcqa-llms-evaluations/2026-01-01T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/prefilling-attack/2026-01-01T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/authenticity/2025-12-17T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/deepfake-detection/2025-12-17T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/multimedia-forensics/2025-12-17T00:00:00+00:00weeklyhttps://seppia978.github.io/publication/2512-arxiv-authentic_multimedia_detection/2025-12-17T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/robustness/2025-12-17T00:00:00+00:00weeklyhttps://seppia978.github.io/event/mbzuai_2511/2025-11-20T14:00:00+00:00weeklyhttps://seppia978.github.io/tags/abu-dhabi-uae/2025-10-20T00:00:00+00:00weeklyhttps://seppia978.github.io/post/2025-10-20T00:00:00+00:00weeklyhttps://seppia978.github.io/post/postdoc-at-mbzuai/2025-10-20T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/mbzuai/2025-10-20T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/ai/2025-10-01T00:00:00+00:00weeklyhttps://seppia978.github.io/post/intro/2025-10-01T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/intro/2025-10-01T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/phd/2025-10-01T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/samuele-poppi/2025-10-01T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/llms/2025-09-22T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/machine-learning-security/2025-09-22T00:00:00+00:00weeklyhttps://seppia978.github.io/publication/2509-iclr-multi-k-wm/2025-09-22T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/watermarking/2025-09-22T00:00:00+00:00weeklyhttps://seppia978.github.io/event/normale_pisa_2505/2025-05-27T14:00:00+00:00weeklyhttps://seppia978.github.io/tags/transformers/2024-12-01T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/trustworthy-ai/2024-12-01T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/unlearning/2024-12-01T00:00:00+00:00weeklyhttps://seppia978.github.io/publication/2412-icpr-unlearning_lora/2024-12-01T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/interpretations-of-multilingual-attacks/2024-10-01T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/multilingual-llm-jailbreaking/2024-10-01T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/multilingual-llms/2024-10-01T00:00:00+00:00weeklyhttps://seppia978.github.io/publication/2504-fnaacl-sil_sir/2024-10-01T00:00:00+00:00weeklyhttps://seppia978.github.io/publication_types/article-journal/2024-06-19T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/explainable-ai/2024-06-19T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/image-classification/2024-06-19T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/machine-unlearning/2024-06-19T00:00:00+00:00weeklyhttps://seppia978.github.io/publication/2406-ieee-multiclass_unlearning/2024-06-19T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/vision-transformers/2024-06-19T00:00:00+00:00weeklyhttps://seppia978.github.io/projects/2024-05-19T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/harmful-content-generation-mitigation/2023-11-27T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/llama-models/2023-11-27T00:00:00+00:00weeklyhttps://seppia978.github.io/publication/2409-eccv-safe_clip/2023-11-27T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/stable-diffusion/2023-11-27T00:00:00+00:00weeklyhttps://seppia978.github.io/experience/2023-10-24T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/clip/2023-09-05T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/computer-vision/2023-09-05T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/deepfakes/2023-09-05T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/dino/2023-09-05T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/image-transformations/2023-09-05T00:00:00+00:00weeklyhttps://seppia978.github.io/publication/2309-iciap-explainable_navigation/2023-09-05T00:00:00+00:00weeklyhttps://seppia978.github.io/publication/2309-iciap-deepfake/2023-09-05T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/vision-and-language/2023-09-05T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/visual-navigation/2023-09-05T00:00:00+00:00weeklyhttps://seppia978.github.io/2022-10-24T00:00:00+00:00weeklyhttps://seppia978.github.io/tags/class-activation-mapping/2021-06-01T00:00:00+00:00weeklyhttps://seppia978.github.io/publication/2106-cvprw-adcc_xai/2021-06-01T00:00:00+00:00weekly