Post-quantum verifiable joint-watermarking for multiparty artificial intelligence models

1 . A computer-implemented method for generating watermarked digital content using generative artificial intelligence (GAI), the method comprising:

receiving a first prompt;

providing a first context window as a hash value by processing a set of previously received tokens through a hash function;

determining a first split parameter using a first key of a first party and a second key of a second party, the first key and the second key generated by a distributed key generation (DKG) process using a channel key, the first split parameter being used to assign tokens to one of a green list and a red list;

providing a first green list of tokens and a first red list of tokens using the first split parameter;

determining a first watermark probability distribution that is biased towards tokens in the first green list of tokens over tokens in the first red list of tokens, such that more tokens are selected from the green list of tokens than are selected from the red list of tokens in generating a watermark; and

returning first watermarked digital content that is generated using the first watermark probability distribution.

2 . The computer-implemented method of claim 1 , further comprising executing a process to establish remote ownership of a trusted execution environment (TEE) by the second party, the TEE being hosted by the first party, the channel key being generated through execution of the process to establish remote ownership of the TEE.

3 . The computer-implemented method of claim 1 , wherein determining a first split parameter comprises:

generating, by the first party, a first split value based on the first context window and a first regulator key; and

generating, by the second party, a second split value based on the first context window and a second regulator key, the first split parameter comprising an aggregation of the first split value and the second split value.

4 . The computer-implemented method of claim 3 , wherein the first split value is generated using a pseudo-random function (PRF) of the first key and the second split value is generated using a PRF of the second key.

5 . The computer-implemented method of claim 3 , wherein the first regulator key is transmitted to the first party by a verifier system and the second regulator key is transmitted to the second party by the verifier system.

6 . The computer-implemented method of claim 1 , wherein the first party determines the first context window and transmits the first context window to the second party, the second party determining a portion of the first split parameter using the first context window.

7 . The computer-implemented method of claim 1 , further comprising verifying, by a verifier system, the first watermarked digital content based on a first verification value determined using first key provided from the first party and a second verification value received from the second party.

8 . The computer-implemented method of claim 1 , further comprising:

receiving a second prompt;

determining a second split parameter using the first key of the first party and the second key of the second party;

providing a second green list of tokens and a second red list of tokens using the second split parameter;

determining a second watermark probability distribution that favors tokens in the second green list of tokens over tokens in the second red list of tokens; and

returning second watermarked digital content that is generated using the second watermark probability distribution.

9 . The computer-implemented method of claim 8 , wherein the first watermark probability distribution is different from the second watermark probability distribution.

10 . The computer-implemented method of claim 8 , further comprising providing a second context window at least partially based on a set of previously received tokens comprising tokens of the first prompt, the second split parameter being determined using the second context window.