【摘要】 可证安全隐写追求由来已久，但是由于缺乏可精确采样的载体，导致可证安全隐写的发展沉寂多年，生成模型的快速发展和生成数据的广泛传播给可证安全隐写带来新的技术手段和伪装环境。文章首先从隐写的经验安全与可证安全的区别出发，引出了可证安全隐写的定义；然后介绍了可证安全隐写的理论与经典构造，并给出了生成式人工智能环境下可证安全隐写的现实构造，在总结基于样本索引的可证安全隐写方法的基础上，介绍了基于分布副本索引的新型隐写构造；最后，文章展望了可证安全隐写的发展趋势：包括公钥隐写和协议，以及在性能无损生成数据水印方面的应用。更多还原

【Abstract】 The pursuit of provable secure steganography has a long history, but there is the lack of accurate sampling carrier, which results in the stagnation of provable secure steganography many years. The rapid development of generative models and the widespread dissemination of generated data bring new technical means and camouflage environments to the provably secure steganography. Firstly, the definition of the provably secure steganography is derived from the difference between empirical security and provable security of steganography. Secondly, the theory and classic construction of the provably secure steganography are introduced, and the realistic construction of the provably secure steganography under the environment of generative artificial intelligence is given, after summarizing the sample-indexes-based provably secure steganography method, a novel steganographic constructions based on distributed replica indexes are introduced. Finally, the development trend of provably secure steganography is prospected, including to the public key steganography and protocols, applications in generating data watermarks without loss of performance.更多还原