IP Library › Granted Patent US 12,651,078
Granted Patent B2
US 12,651,078 · App. 19/363,568 · Granted Jun 9, 2026

Monitoring and controlling communications between autonomous agents

Inventors: Manjit Rajaretnam (Irving, TX); Sofia Rahman (New York, NY); William Cameron (Jacksonville, FL); Imir Arifi (Nashville, TN); James Myers (New York, NY)
Assignee: Citibank, N.A.
G06F21/577G06F21/552
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 12,651,078
App. No.
19/363,568
Filed
Oct 20, 2025
Granted
Jun 9, 2026
Kind
B2
Art Unit
2435
USPC
726/22
Abstract

Systems, methods, and devices for monitoring and controlling communications between autonomous agents are disclosed. The system monitors real-time communications between autonomous agents, intercepting and recording each communication. Communications are translated to a standardized language and processed through communication protocol filters that evaluate compliance with predefined operational policies. The system parses each translated communication to identify policy violations. When violations are detected, the system modifies communications to ensure compliance with operational policies. All communications and modifications are recorded via distributed ledger technology for audit and accountability purposes. This approach enables comprehensive oversight of autonomous agent interactions while maintaining tamper-proof records of all monitoring and control activities.

Claims (38)

1 . One or more non-transitory, computer-readable storage medium comprising instructions recorded thereon, wherein the instructions, when executed by at least one data processor of a system, cause the system to:

monitor, in real time, a series of communications between a plurality of autonomous agents operating within an operational environment, wherein the plurality of autonomous agents are organized into a plurality of agent meshes based on functional domains of the plurality of autonomous agents, the series of communications comprising intra-mesh communications within individual agent meshes and inter-mesh communications across agent mesh boundaries of some of the individual agent meshes, and wherein monitoring the series of communications comprises intercepting and recording each communication of the series of communications;

translate each communication in the series of communications to a standardized language;

implement, for each communication translated to the standardized language, at least one communication protocol filter, wherein the at least one communication protocol filter is configured to evaluate each communication based on predefined operational policies, and wherein the predefined operational policies comprise domain-specific policies associated with respective individual agent meshes of the plurality of agent meshes;

using the at least one communication protocol filter, parse each communication translated to the standardized language to determine whether each communication violates at least one relevant domain-specific policy of the domain-specific policies;

based on determining that a particular communication translated to the standardized language violates the at least one relevant domain-specific policy, perform one or more modifications to the particular communication to render the particular communication compliant with the at least one relevant domain-specific policy; and

execute a smart contract to record, via a distributed ledger, the particular communication and the one or more modifications to the particular communication.

2 . The one or more non-transitory, computer-readable storage medium of claim 1 , wherein the instructions for monitoring the series of communications further cause the system to employ a transformer-based language model configured to automatically detect and translate communications that utilize novel or evolving AI agent languages.

3 . The one or more non-transitory, computer-readable storage medium of claim 1 , wherein the at least one communication protocol filter comprises hybrid violation detection using both a rule-based filter encoded with a declarative policy language and a machine learning-based model trained to score violation risk.

4 . The one or more non-transitory, computer-readable storage medium of claim 1 , wherein the instructions for parsing each communication further cause the system to apply semantic analysis through natural language processing models for policy violation assessment.

5 . The one or more non-transitory, computer-readable storage medium of claim 1 , wherein the instructions for parsing each communication further cause the system to perform anomaly detection on message frequency, entropy, or known abuse patterns.

6 . The one or more non-transitory, computer-readable storage medium of claim 1 , wherein the instructions for performing the one or more modifications to the particular communication further cause the system to block or quarantine the particular communication.

7 . A method comprising:

monitoring, in real time, a series of communications between a plurality of autonomous AI agents operating within an operational computing environment, wherein the plurality of autonomous AI agents are organized into a plurality of agent meshes based on functional domains of the plurality of autonomous AI agents, the series of communications comprising intra-mesh communications within agent meshes and inter-mesh communications across agent mesh boundaries;

implementing, for the series of communications, at least one communication protocol filter, wherein the at least one communication protocol filter is configured to evaluate the series of communications based on predefined operational policies, and wherein the predefined operational policies comprise domain-specific policies associated with respective agent meshes of the plurality of agent meshes;

determining, using the at least one communication protocol filter, whether a particular communication of the series of communications violates at least one domain-specific policy of the domain-specific policies;

based on determining that the particular communication violates the at least one domain-specific policy, performing one or more modifications to the particular communication to render the particular communication compliant with the at least one domain-specific policy; and

recording, via a distributed ledger, the particular communication and the one or more modifications to the particular communication.

8 . The method of claim 7 , wherein monitoring the series of communications further comprises employing a transformer-based language model configured to automatically detect and translate communications that utilize novel or evolving AI agent languages.

9 . The method of claim 7 , wherein the at least one communication protocol filter comprises hybrid violation detection using both a rule-based filter encoded with a declarative policy language and a machine learning-based model trained to score violation risk.

10 . The method of claim 7 , wherein determining whether the particular communication of the series of communications violates the at least one domain-specific policy further comprises applying semantic analysis through natural language processing models for policy violation assessment.

11 . The method of claim 7 , wherein determining whether the particular communication of the series of communications violates the at least one domain-specific policy further comprises performing anomaly detection on message frequency, entropy, or known abuse patterns.

12 . The method of claim 7 , wherein performing the one or more modifications to the particular communication further comprises blocking or quarantining the particular communication.

13 . The method of claim 7 , wherein monitoring the series of communications further comprises intercepting and recording the series of communications.

14 . A system comprising:

a storage device; and

one or more processors communicatively coupled to the storage device storing instructions thereon, that cause the one or more processors to:

monitor, in real time, a series of communications between a plurality of autonomous AI agents operating within an operational computing environment, wherein the plurality of autonomous AI agents are organized into a plurality of agent meshes based on functional domains of the plurality of autonomous agents;

implement, for the series of communications, at least one communication protocol filter, wherein the at least one communication protocol filter is configured to evaluate the series of communications based on predefined operational policies, and wherein the predefined operational policies comprise domain-specific policies associated with respective agent meshes of the plurality of agent meshes;

determine, using the at least one communication protocol filter, whether a particular communication of the series of communications violates at least one domain-specific policy of the domain-specific policies;

based on determining that the particular communication violates the at least one domain-specific policy, perform one or more modifications to the particular communication to render the particular communication compliant with the at least one domain-specific policy; and

record, via a distributed ledger, the particular communication and the one or more modifications to the particular communication.

15 . The system of claim 14 , wherein the instructions for monitoring the series of communications further cause the one or more processors to employ a transformer-based language model configured to automatically detect and translate communications that utilize novel or evolving AI agent languages.

16 . The system of claim 14 , wherein the at least one communication protocol filter comprises hybrid violation detection using both a rule-based filter encoded with a declarative policy language and a machine learning-based model trained to score violation risk.

17 . The system of claim 14 , wherein the instructions for determining whether the particular communication of the series of communications violates the at least one domain-specific policy further cause the one or more processors to apply semantic analysis through natural language processing models for policy violation assessment.

18 . The system of claim 14 , wherein the instructions for determining whether the particular communication of the series of communications violates the at least one domain-specific policy further cause the one or more processors to perform anomaly detection on message frequency, entropy, or known abuse patterns.

19 . The system of claim 14 , wherein the instructions for performing the one or more modifications to the particular communication further cause the one or more processors to block or quarantine the particular communication.

20 . The system of claim 14 , wherein the instructions for monitoring the series of communications further cause the one or more processors to intercept and record the series of communications.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Oct 27, 2025
From: RAJARETNAM, MANJIT; RAHMAN, SOFIA; CAMERON, WILLIAM; ARIFI, IMIR; MYERS, JAMES
To: CITIBANK, N.A.
Reel/Frame 072689/0895 →
Continuity (22)
Continuation In Part 19283194 · Jul 28, 2025
Continuation In Part 19182585 · Apr 18, 2025
Continuation 18947102 · Nov 14, 2024
Continuation In Part 18653858 · May 2, 2024
Continuation In Part 18637362 · Apr 16, 2024
Continuation In Part 18782019 · Jul 23, 2024
Continuation In Part 18771876 · Jul 12, 2024
Continuation In Part 18661532 · May 10, 2024
Continuation In Part 18661519 · May 10, 2024
Continuation In Part 18633293 · Apr 11, 2024
Continuation In Part 18739111 · Jun 10, 2024
Continuation In Part 18607141 · Mar 15, 2024
Continuation In Part 18399422 · Dec 28, 2023
Continuation 18327040 · May 31, 2023
Continuation In Part 18114194 · Feb 24, 2023
Continuation In Part 18098895 · Jan 19, 2023
Continuation In Part 18951366 · Nov 18, 2024
Continuation 18624409 · Apr 2, 2024
Continuation In Part 19195642 · Apr 30, 2025
Continuation In Part 18762362 · Jul 2, 2024
Continuation 18624409 · Apr 2, 2024
Related Publication 20260044610A1 · Feb 12, 2026
References Cited (32)
US 8959633B1 · Dokey et al. · 2015 [cited by applicant]
US 10430581B2 · Lamothe-brassard · 2019 [cited by applicant]
US 10671726B1 · Paithane et al. · 2020 [cited by applicant]
US 10902117B1 · Singh et al. · 2021 [cited by applicant]
US 10902119B1 · Vashisht et al. · 2021 [cited by applicant]
US 11106801B1 · Levine et al. · 2021 [cited by applicant]
US 11227047B1 · Vashisht et al. · 2022 [cited by applicant]
US 11810185B2 · Joglekar et al. · 2023 [cited by applicant]
US 12130914B1 · Bushner et al. · 2024 [cited by applicant]
US 12299134B1 · Li et al. · 2025 [cited by applicant]
US 12375573B1 · Nanduri et al. · 2025 [cited by applicant]
US 12450363B2 · Ohayon et al. · 2025 [cited by applicant]
US 20070240215A1 · Flores et al. · 2007 [cited by applicant]
US 20160359890A1 · Deen et al. · 2016 [cited by applicant]
US 20170279839A1 · Vasseur et al. · 2017 [cited by applicant]
US 20220124013A1 · Chitalia et al. · 2022 [cited by applicant]
US 20220191224A1 · Ståhlberg et al. · 2022 [cited by applicant]
US 20230123322A1 · Cella · 2023 [cited by examiner]
US 20230208855A1 · Sheriff et al. · 2023 [cited by applicant]
US 20230208858A1 · Mishra et al. · 2023 [cited by applicant]
US 20230261857A1 · Baldwin et al. · 2023 [cited by applicant]
US 20230379350A1 · Barton et al. · 2023 [cited by applicant]
US 20230396435A1 · Baldwin et al. · 2023 [cited by applicant]
US 20230409756A1 · Baldwin et al. · 2023 [cited by applicant]
US 20240202323A1 · Hoogerbrugge et al. · 2024 [cited by applicant]
US 20240333737A1 · Shahrivar et al. · 2024 [cited by applicant]
US 20240333768A1 · Shahrivar et al. · 2024 [cited by applicant]
US 20240394401A1 · Agrawal et al. · 2024 [cited by applicant]
US 20240403413A1 · Cohen · 2024 [cited by examiner]
US 20250217484A1 · Pachisia et al. · 2025 [cited by applicant]
US 20250301011A1 · Niemelä · 2025 [cited by applicant]
Dong, Y., et al., “Building Guardrails for Large Language Models,” https://ar5iv.labs.arxiv.org/html/2402.01822v1, published May 29, 2024, 20 pages. [cited by applicant]