IP Library Granted Patent US 12,574,213
Granted Patent B2
US 12,574,213 · App. 18/813,517 · Granted Mar 10, 2026

Key-based generative artificial intelligence model content generation

Inventors: Mayank Kumar Singh (Tokyo, JP); Naoya Takahashi (Zurich, CH)
Assignee: Sony Group Corporation
H04L9/0819H04L9/3297
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,574,213
App. No.
18/813,517
Granted
Mar 10, 2026
Kind
B2
Abstract

In some implementations, a system may receive an authentication key associated with authenticating use of a generative AI model. The system may receive an input prompt. The system may generate, using the generative AI model, an output based on the input prompt. The authentication key is embedded within the output. A quality of the output is higher when the authentication key is valid than when the authentication key is not valid.

Claims (84)

1 . A method for generating an output of a generative artificial intelligence (AI) model using an authentication key, the method comprising:

receiving, by a device, an authentication key associated with authenticating use of the generative AI model;

receiving, by the device, an input prompt;

generating, by the device and using the generative AI model, an output based on the input prompt,

wherein generating the output includes embedding the authentication key within the output, and

wherein a quality of the output is higher when the authentication key is valid than when the authentication key is not valid.

2 . The method of claim 1 , wherein the use of the model is authenticated by receiving the authentication key after access to the generative AI model has been provided.

3 . The method of claim 1 , wherein the output includes at least one of:

an image output,

an audio output,

a text output,

a video output,

a multimodal output, or

a synthetic content output.

4 . The method of claim 1 , further comprising:

receiving, by the device, metadata,

wherein generating the output includes embedding the metadata within the output.

5 . The method of claim 4 , wherein the metadata includes at least one of:

a generative AI model name,

a generative AI model version, or

a timestamp of generation of the output.

6 . The method of claim 1 , wherein the quality of the output is altered from a higher quality to a lower quality when the authentication key is not valid.

7 . The method of claim 1 , wherein access to the generative AI model is at least one of:

a server-side access, or

a client-side access.

8 . A system for generating an output of a generative artificial intelligence (AI) model using an authentication key, the system comprising:

one or more memories; and

one or more processors, communicably coupled to the one or more memories, configured to:

receive an authentication key associated with authenticating use of the generative AI model;

receive an input prompt;

generate, using the generative AI model, an output based on the input prompt,

wherein generating the output includes embedding the authentication key within the output, and

wherein a quality of the output is higher when the authentication key is valid than when the authentication key is not valid.

9 . The system of claim 8 , wherein the use of the model is authenticated by receiving the authentication key after access to the generative AI model has been provided.

10 . The system of claim 8 , wherein the output includes at least one of:

an image output,

an audio output,

a text output,

a video output,

a multimodal output, or

a synthetic content output.

11 . The system of claim 8 , wherein the one or more processors are further configured to:

receive metadata,

wherein generating the output includes embedding the metadata within the output.

12 . The system of claim 11 , wherein the metadata includes at least one of:

a generative AI model name,

a generative AI model version, or

a timestamp of generation of the output.

13 . The system of claim 8 , wherein the quality of the output is altered from a higher quality to a lower quality when the authentication key is not valid.

14 . The system of claim 8 , wherein access to the generative AI model is at least one of:

a server-side access, or

a client-side access.

15 . A method for generating an output of a generative artificial intelligence (AI) model using an authentication key, the method comprising:

receiving, by a device, an authentication key;

authenticating, by the device and based on the authentication key, use of the generative AI model by a user prior to the user gaining access to the generative AI model;

receiving, by a device, an input prompt;

generating, by the device and using the generative AI model, an output based on the input prompt,

wherein generating the output includes embedding the authentication key within the output, and

wherein a quality of the output is higher when the authentication key is valid than when the authentication key is not valid.

16 . The method of claim 15 , wherein the output includes at least one of:

an image output,

an audio output,

a text output,

a video output,

a multimodal output, or

a synthetic content output.

17 . The method of claim 15 , wherein the quality of the output is altered from a higher quality to a lower quality when the authentication key is not valid.

18 . A system for generating an output of a generative artificial intelligence (AI) model using an authentication key, the system comprising:

one or more memories; and

one or more processors, communicably coupled to the one or more memories, configured to:

receive an authentication key;

authenticate, based on the authentication key, use of the generative AI model by a user prior to the user gaining access to the generative AI model;

receive an input prompt;

generate, using the generative AI model, an output based on the input prompt,

wherein generating the output includes embedding the authentication key within the output, and

wherein a quality of the output is higher when the authentication key is valid than when the authentication key is not valid.

19 . The system of claim 18 , wherein the output includes at least one of:

an image output,

an audio output,

a text output,

a video output,

a multimodal output, or

a synthetic content output.

20 . The system of claim 18 , wherein the quality of the output is altered from a higher quality to a lower quality when the authentication key is not valid.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Oct 25, 2024
From: SINGH, MAYANK KUMAR; TAKAHASHI, NAOYA
To: SONY GROUP CORPORATION
Reel/Frame 069024/0688 →
Continuity (1)
Related Publication 20260058796A1 · Feb 26, 2026
References Cited (13)
US 12236202B1 · DeWeese · 2025 [cited by examiner]
US 12322402B2 · Drolet · 2025 [cited by examiner]
US 20130004033A1 · Trugenberger · 2013 [cited by examiner]
US 20210319090A1 · Demir · 2021 [cited by examiner]
US 20210319240A1 · Demir et al. · 2021 [cited by applicant]
US 20240152544A1 · Aykut et al. · 2024 [cited by applicant]
US 20250209200A1 · Paes · 2025 [cited by examiner]
Tang, Nan, et al. “VerifAI: verified generative AI.” arXiv preprint arXiv:2307.02796 (2023). (Year: 2023). [cited by examiner]
Tihanyi, Norbert, et al. “The formai dataset: Generative ai in software security through the lens of formal verification.” Proceedings of the 19th International Conference on Predictive Models and Data Analytics in Soft… [cited by examiner]
Plevris Vagelis et al: “Chatbots Put to the Test in Math and Logic Problems: A Comparison and Assessment of ChatGPT-3.5, ChatGPT-4, and Google Bard”, Predictive Society and Data Analytics Lab, Faculty of Information Tec… [cited by applicant]
Yepeng Liu et al: “Adaptive Text Watermark for Large Language Models”, arxiv.org, Cornell University Library, 201 Olin Library Cornell University Ithaca, NY 14853, Jun. 9, 2024 (Jun. 9, 2024), XP091781486. [cited by applicant]
Wenjie Qu et al: “Provably Robust Muiti-bit Watermarking for AI-generated Text via Error Correction Code”, arxiv.org, Cornell University Library, 201 Olin Library Cornell University Ithaca, NY 14853, Apr. 16, 2024 (Apr.… [cited by applicant]
Kiyoon Yoo et al: “Advancing Beyond Identification: Multi-bit Watermark for Large Language Models”, arxiv.org, Cornell University Library, 201 Olin Library Cornell University Ithaca, NY 14853, Sep, 27, 2023 (Sep. 27, 20… [cited by applicant]