ARTIFICIAL INTELLIGENCE SAFETY AND ROBUSTNESS AGAINST ADVERSARIAL ATTACKS
Keywords:
AI safety, adversarial attacks, robustness, adversarial examples, red teamingAbstract
This scientific article provides a comprehensive analysis of the safety of
artificial intelligence (AI) systems and their robustness against adversarial attacks. As AI is
increasingly used in critical fields (medicine, transport, security), the safety and reliability of these
systems become extremely important. The research scientifically substantiates the nature of
adversarial attacks — specially crafted inputs designed to deceive AI — methods of defending
against them, and the broader problems of AI safety. The article examines the vulnerabilities of AI
systems, the methods of "red teaming" (testing for weaknesses), and ways to build reliable and
robust systems. The scientific novelty of the article lies in demonstrating that AI safety is a
fundamental condition for the trustworthy use of these technologies. As a result of the analyses,
recommendations are developed regarding the development of safe AI systems
References
1. Szegedy, C., et al. (2014). "Intriguing properties of neural networks." ICLR.
2. Goodfellow, I., et al. (2015). "Explaining and Harnessing Adversarial Examples." ICLR.
3. Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep Learning. MIT Press.
4. Madry, A., et al. (2018). "Towards Deep Learning Models Resistant to Adversarial Attacks."
ICLR.
5. Amodei, D., et al. (2016). "Concrete Problems in AI Safety." arXiv preprint.
6. Carlini, N., & Wagner, D. (2017). "Towards Evaluating the Robustness of Neural Networks."
IEEE Symposium on Security and Privacy.
7. Papernot, N., et al. (2016). "The Limitations of Deep Learning in Adversarial Settings." IEEE
European Symposium on Security and Privacy.
8. Eykholt, K., et al. (2018). "Robust Physical-World Attacks on Deep Learning Visual
Classification." CVPR.
9. Vaswani, A., et al. (2017). "Attention is All You Need." Advances in Neural Information
Processing Systems.
10. Russell, S. (2019). Human Compatible: Artificial Intelligence and the Problem of Control.
Viking.
11. Kurakin, A., et al. (2017). "Adversarial examples in the physical world." ICLR Workshop.
12. Russell, S., & Norvig, P. (2020). Artificial Intelligence: A Modern Approach (4th ed.). Pearson.
13. Hendrycks, D., et al. (2021). "Unsolved Problems in ML Safety." arXiv preprint.
14. Biggio, B., & Roli, F. (2018). "Wild patterns: Ten years after the rise of adversarial machine
learning." Pattern Recognition.
15. Bostrom, N. (2014). Superintelligence: Paths, Dangers, Strategies. Oxford University Press.
16. Decree of the President of the Republic of Uzbekistan "On the approval of the strategy 'Digital
Uzbekistan – 2030'" (October 5, 2020).
17. UNESCO. (2021). Recommendation on the Ethics of Artificial Intelligence.
18. Floridi, L. (2019). The Ethics of Artificial Intelligence: Principles, Challenges, and
Opportunities. Oxford University Press.
