>>> bs = AES.block_size
>>> pad = lambda s: s + (bs - len(s) % bs) * chr(bs - len(s) % bs)
>>> key = "password1234".encode() # to bytes
>>> aes_1 = AES.new(md5(key).digest(), AES.MODE_CFB, bytes([random.randint(0,0 xFF) for i in range(16)]))
>>> aes_2 = AES.new(md5(key).digest(), AES.MODE_CFB, bytes([random.randint(0,0 xFF) for i in range(16)]))
>>> d_text = pad("texttexttexttext blablablah")
>>> e_text = aes_1.encrypt(d_text)
b texttexttexttext blablablah\x05\x05\x05\x05\x05'
Question: why aes_1 that the text was encrypted, couldn't the text be transcribed, and aes_2 could? All but the initial vector is the same, ie that would decipher the text IV you need to randomize? Why, then, nowhere it is written? Or so difficult to read/googled?
UPD: Pogorel more, read about the modes in detail, I understood why can't rasshifrovat, tried again - now aes_1 and aes_2 produces garbage instead of text. CANT?