Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claim Rejections - 35 USC § 102
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claim(s) 10-11, and 14-18 are rejected under 35 U.S.C. 102(a)(2) as being anticipated by McGrew et al (2024/0333765).
For 10, McGrew teaches method for configuring a honeypot (abstract), comprising the following steps: implementing a honeypot (McGrew teaches of generated honeypot schemes as McGrew teaches in par.20); conducting at least one attack on the honeypot using a large language model (attack vector generator 222 converts the subgraph 220 of detected malware identified during penetration testing into a plurality of attack vectors utilization of Large Language Model (LLM)-generated honeypot schemes that involve the creation of deceptive documents as McGrew teaches in par.42 and 75); ascertaining an assessment of the at least one attack (McGrew teaches that continuously monitoring for interactions initiated by an interacting party with one or more components of the generated deceptive information, where the interaction is identified as a potential threat to the network as McGrew teaches in par.23); and configuring the honeypot depending on the assessment of the at least one attack (McGrew teaches that in response to the detection of an interaction identified as the potential threat, the method further includes generating one or more remedial measures and policies for the network based on the extracted interaction data, enhancing network security as McGrew teaches in par.27 and 28-32, 78-81).
For 11, McGrew further teaches wherein, for ascertaining the assessment of the at least one attack, an assessment with regard to a number of interactions of the large language model with the honeypot takes place (McGrew teaches that monitoring includes generating one or more contextual labels in accordance with contextual data related to the interactions to distinguish and identify threatening interactions from non-threatening interactions, and retraining the LLM with the one or more contextual labels improving an accuracy level of potential threat detection and effectiveness of the honeypots as McGrew teaches in par.29 and 31).
For 14, McGrew further teaches training or retraining the large language model based on examples of exploiting known vulnerabilities (McGrew teaches that retraining the LLM with the one or more contextual labels improving an accuracy level of potential threat detection and effectiveness of the honeypots as McGrew teaches in par.29).
For 15, McGrew further teaches implementing the honeypot for each of a plurality of configurations (McGrew teaches configuration generator 226 then generates a policy 228 for the prompt generator for implementing the honeypot as McGrew teaches in par.78); conducting, for each configuration, at least one attack on the honeypot using the large language model (McGrew teaches of applying multiple attacks using the large language model as McGrew teaches in par.22-23 and 75); ascertaining an assessment of the at least one attack for each configuration; selecting a configuration of the honeypot from the plurality of configurations that provides a best assessment relative of assessments of the others of the plurality of configurations (McGrew teaches that continuously monitoring for interactions initiated by an interacting party with one or more components of the generated deceptive information, where the interaction is identified as a potential threat to the network as McGrew teaches in par.23); and configuring the honeypot according to the selected configuration (McGrew teaches that in response to the detection of an interaction identified as the potential threat, the method further includes generating one or more remedial measures and policies for the network based on the extracted interaction data, enhancing network security as McGrew teaches in par.27 and 28).
For 16, McGrew teaches A honeypot configuration device (abstract) comprising at least one processor, the at least one processor configured (McGrew teaches of generated honeypot schemes as McGrew teaches in par.20-23);; conduct at least one attack on the honeypot using a large language model (attack vector generator 222 converts the subgraph 220 of detected malware identified during penetration testing into a plurality of attack vectors utilization of Large Language Model (LLM)-generated honeypot schemes that involve the creation of deceptive documents as McGrew teaches in par.42 and 75); ascertain an assessment of the at least one attack (McGrew teaches that continuously monitoring for interactions initiated by an interacting party with one or more components of the generated deceptive information, where the interaction is identified as a potential threat to the network as McGrew teaches in par.23); and configure the honeypot depending on the assessment of the at least one attack (McGrew teaches that in response to the detection of an interaction identified as the potential threat, the method further includes generating one or more remedial measures and policies for the network based on the extracted interaction data, enhancing network security as McGrew teaches in par.27 and 28-32, 78-81).
For 17, McGrew teaches A non-transitory computer-readable medium on which are stored commands configuring a honeypot, the commands, when executed by a processor (par.32), causing the processor to perform the following steps: implementing a honeypot (McGrew teaches of generated honeypot schemes as McGrew teaches in par.20); conducting at least one attack on the honeypot using a large language model (attack vector generator 222 converts the subgraph 220 of detected malware identified during penetration testing into a plurality of attack vectors utilization of Large Language Model (LLM)-generated honeypot schemes that involve the creation of deceptive documents as McGrew teaches in par.42 and 75); ascertaining an assessment of the at least one attack (McGrew teaches that continuously monitoring for interactions initiated by an interacting party with one or more components of the generated deceptive information, where the interaction is identified as a potential threat to the network as McGrew teaches in par.23); and configuring the honeypot depending on the assessment of the at least one attack (McGrew teaches that in response to the detection of an interaction identified as the potential threat, the method further includes generating one or more remedial measures and policies for the network based on the extracted interaction data, enhancing network security as McGrew teaches in par.27 and 28-32, 78-81).
For 18, McGrew further teaches wherein the assessment takes into account quality of functionalities of a target system implemented by the honeypot (McGrew teaches that where the continuously monitoring includes generating one or more contextual labels in accordance with contextual data related to the interactions to distinguish and identify threatening interactions from non-threatening interactions, and retraining the LLM with the one or more contextual labels improving an accuracy level of potential threat detection and effectiveness of the honeypots as McGrew teaches in par.27-32).
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claim(s) 12 is/are rejected under 35 U.S.C. 103 as being unpatentable over McGrew et al (2024/0333765) in views of Lu et al (CN 116886446).
McGrew teaches all the limitations as previously set forth except for wherein the configuring of the honeypot depending on the assessment of the at least one attack includes ascertaining whether the assessment of the at least one attack is above a specified threshold value and, in response to the assessment of the at least one attack not being above the specified threshold value, changing a behavior of the honeypot to a state in which the honeypot was when the large language model no longer made progress on the attack.
Lu teaches that wherein the configuring of the honeypot depending on the assessment of the at least one attack includes ascertaining whether the assessment of the at least one attack is above a specified threshold value and, in response to the assessment of the at least one attack not being above the specified threshold value, changing a behavior of the honeypot to a state in which the honeypot was when the large language model no longer made progress on the attack (Lu teaches detection behaviour of the target large language model triggered by the access operation of the target client in the target time window that the target large language model will be triggered when the access frequency of the same client exceeds the set access frequency threshold and under the condition that the access frequency threshold is not exceeded as Lu teaches in par.18). It would have been obvious to one ordinary skill in the art before effective filling date to modify McGrew to include specified threshold value as taught and suggested by Lu for the purpose of collecting the behavior characteristic parameter corresponding to each triggered preset detection behavior; obtaining all target interaction data between the target client and the target large language model in the target time window and determining whether the target client is performing the automatic attack (Lu, abstract).
Claim(s) 13 is/are rejected under 35 U.S.C. 103 as being unpatentable over McGrew et al (2024/0333765) in views of Kawasaki et al (2018/0375897).
McGrew teaches all the limitations as previously set forth and McGrew further teaches that wherein the conducting the at least one attack on the honeypot includes generating, by the large language model (McGrew teaches in par.22-23). However, McGrew fails to teach that at least one input for a command line interface that the honeypot simulates, and feeding the at least one generated input into the simulated command line interface.
Kawasaki teaches at least one input for a command line interface that the honeypot simulates, and feeding the at least one generated input into the simulated command line interface (Kawasaki teaches honey pots responding to commands and the commands are entered via interface as Kawasaki teaches in par.16-17, 20 and 32). It would have been obvious to one ordinary skill in the art before effective filling date to modify McGrew to include command line interface as taught and suggested by Kawasaki for the purpose of spares the valuable assets, and the honeypots may be monitored for attacks, so that computer and network administrators may be alerted of attacks in progress (Kawasaki, par.16).
Claim(s) 19-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over McGrew et al (2024/0333765) in views of McDonald et al (2025/0168186).
For claim 19, McGrew teaches all the limitations as previously set forth and McGrew further teaches that wherein conducting the at least one attack on the honeypot comprises generating, by the large language model (McGrew teaches in par.22-23). However, McGrew fails to teach that at least one input for a command line interface that the honeypot simulates, and feeding the at least one generated input into the simulated command line interface, wherein a response of the command line interface is fed back into the large language model as a prompt SO that the large language model can generate a new input.
McDonald teaches, similar system, at least one input for a command line interface that the honeypot simulates, and feeding the at least one generated input into the simulated command line interface, wherein a response of the command line interface is fed back into the large language model as a prompt SO that the large language model can generate a new input (McDonald teaches that a plurality of a components (e.g., generative AI model 102D, host machine 104D, attacker machine 106D, victim machine 108D, host agent 140D, and attacker agent 150D); host machine operations (e.g., assume the initial situation, at step 110D, prepare the prompt for LLM, at step 118D, summarize the response, at step 124D, determine failure/success/continue, at step 132D, loop until completing the task or failure, at step 116D); and a plurality of communications. At the beginning of a task, an initial situation (e.g., attack profile) of a task can be described and the task target in the prompt to give the generative AI model background information and reset all the VMs in the Hyper-V (e.g., via Powershell Direct). As such, based on the initial situation, the host machine at step 110D can assume the initial situation associated with a task, The host machine 104D, at step 112D resets the attacker machine 106D and, at step 114D, resets the victim machine 108D to initialize the attacker machine 106D and the victim machine 108D. After initialization, the cybersecurity simulation engine can be iteratively implemented to prepare different prompts for the generative AI model (“LLM”) (e.g., loop until completing the task or failure, at step 116D). At step 118D, the host machine 104D prepares the prompt for the LLM 102D. The host machine 104D at step 120D, sends the prompt to the LLM 102; and at step 122D, receives a response to the prompt. At step 124D, the host machine 104D summarizes the response; and at step 126D, assigns commands by writing to a file that is accessible by the attacker machine 106D. In this way, the cybersecurity engine receives a response and shares the response with the attacker machine 106D in a file (e.g., a Kali Linux file), An attacker machine uses a specific format, reasoning, and command information to get accurate information and commands. The specific format, reasoning, and command information can be extracted from the file. The attacker machine can execute commands. As shown, at step 128D, the attacker machine 106D executes commands on the victim machine 108D; and accesses responses from the executed commands, and at step 130D, provides the response to the host machine 104D. In particular, the attacker machine 106D, receives a command response and returns the response to the host machine 104D (e.g., an agent in the host machine) using the same file. The attacker machine 106D can wait for the host machine 104D to decide whether to loop or terminate based on whether it meets the maximum number interactions or complete the tasks. As shown at step 132D, a failure, success, continue determination can be made, and based on step 116D, the procedure can loop until complete the task or failure (or any other predefined termination criteria) as McDonald teaches in par.41-47) (fig.1D). It would have been obvious to one ordinary skill in the art before effective filling date to modify McGrew to include wherein a response of the command line interface is fed back into the large language model as a prompt SO that the large language model can generate a new input as taught and suggested by McDonald for the purpose of providing cybersecurity operations and interfaces via a cybersecurity engine in a security management system. As such, the security management system can be improved based on cybersecurity engine operations that operate to effectively simulate cybersecurity attacks using generative AI models (McDonald, par.6).
For claim 20, McGrew teaches all the limitations as previously set forth and McGrew further teaches that wherein the large language model is used to simulate an attacker attacking a honeypot, (McGrew teaches in par.22-23). However, McGrew fails to teach that wherein the responses of the command line interface are in turn fed into the large language model to continue with the attack.
McDonald further teaches that wherein the responses of the command line interface are in turn fed into the large language model to continue with the attack (McDonald teaches that As shown, at step 128D, the attacker machine 106D executes commands on the victim machine 108D; and accesses responses from the executed commands, and at step 130D, provides the response to the host machine 104D. In particular, the attacker machine 106D, receives a command response and returns the response to the host machine 104D (e.g., an agent in the host machine) using the same file. The attacker machine 106D can wait for the host machine 104D to decide whether to loop or terminate based on whether it meets the maximum number interactions or complete the tasks. As shown at step 132D, a failure, success, continue determination can be made, and based on step 116D, the procedure can loop until complete the task or failure (or any other predefined termination criteria) as McDonald teaches in par.41-47) (fig.1D). It would have been obvious to one ordinary skill in the art before effective filling date to modify McGrew to include wherein the responses of the command line interface are in turn fed into the large language model to continue with the attack as taught and suggested by McDonald for the purpose of providing cybersecurity operations and interfaces via a cybersecurity engine in a security management system. As such, the security management system can be improved based on cybersecurity engine operations that operate to effectively simulate cybersecurity attacks using generative AI models (McDonald, par.6).
Response to Amendments/Arguments
Applicant's arguments filed 07/09/2026 have been fully considered but they are not persuasive.
The Applicant's arguments see on page 2-7 of the remarks have been fully considered but they are not persuasive. The applicant argues in page 2 of remarks, that McGrew fails to disclose "conducting at least one attack on the honeypot using a large language model," as recited in claim 10. However, examiner respectfully disagrees with the applicant because McGrew teaches that enhancing cybersecurity using Large Language Model (LLM)-generated honeypot schemes, the method includes generating a plurality of deceptive information using an LLM, configured to attract and engage potential attackers, where the plurality of deceptive information includes one or more characteristics referencing vulnerabilities of a network, continuously monitoring for interactions initiated by an interacting party with one or more components of the generated deceptive information, where the interaction is identified as a potential threat to the network, in response to detection of an interaction identified as a potential threat, extracting interaction data associated with the interacting party retrieved during the interaction, and retraining the LLM with the interaction data to create more effective honeypots, LLMs offer a strategic advantage in the creation of proactive honeypots designed to apprehend malicious actors attempting to evade detection. Through the utilization of LLMs, organizations can gain the capability to generate honeypots boasting a diverse array of potential attack vectors and anticipated observables. This extensive range empowers organizations to enhance their preparedness against forthcoming attacks and swiftly identify suspicious activities as they unfold in real-time, during web-based attacks, penetration testing can detect tactics such as cross-site scripting (XSS), where attackers inject malicious scripts into web pages, and SQL injection, which involves manipulating databases through improperly sanitized inputs. Additionally, common attack vectors that target operating systems can be exposed by exploiting known vulnerabilities to gain unauthorized access. Examples of such threats include privilege escalation attacks buffer overflow attacks, and the execution of arbitrary code. Therefore, McGrew’s system meets the claim limitation.
The applicant argues in page 4 of remarks, that McGrew fails to disclose "for ascertaining the assessment of the at least one attack, an assessment with regard to a number of interactions of the large language model with the honeypot takes place," as recited in claim 11. However, examiner respectfully disagrees with the applicant because McGrew teaches that generate a plurality of deceptive information using an LLM, configured to attract and engage potential attackers, where the plurality of deceptive information includes one or more characteristics referencing vulnerabilities of the network, continuously monitor for interactions initiated by an interacting party with one or more components of the generated deceptive information, where the interaction is identified as a potential threat to the network, in response to detection of an interaction identified as a potential threat, extract interaction data associated with the interacting party retrieved during the interaction, and retrain the LLM with the interaction data to create more effective honeypots. Therefore, McGrew’s system meets the claim limitation.
The applicant argues in page 4 of remarks, that McGrew fails to disclose "training or retraining the large language model based on examples of exploiting known vulnerabilities," as recited in claim 14. However, examiner respectfully disagrees with the applicant because McGrew teaches that training the LLM by leveraging malware data stored in a storage, where the LLM references the malware data to identify potential threats in a network, The network device also includes one or more processors configured to execute the computer-readable instructions to generate a plurality of deceptive information using an LLM, configured to attract and engage potential attackers, where the plurality of deceptive information includes one or more characteristics referencing vulnerabilities of the network, continuously monitor for interactions initiated by an interacting party with one or more components of the generated deceptive information, where the interaction is identified as a potential threat to the network, in response to detection of an interaction identified as a potential threat, extract interaction data associated with the interacting party retrieved during the interaction, and retrain the LLM with the interaction data to create more effective honeypots. Therefore, McGrew’s system meets the claim limitation.
The applicant argues in page 5 of remarks, that McGrew fails to disclose "implementing the honeypot for each of a plurality of configurations; conducting, for each configuration, at least one attack on the honeypot using the large language model; ascertaining an assessment of the at least one attack for each configuration; selecting a configuration of the honeypot from the plurality of configurations that provides a best assessment relative of assessements of the others of the plurality of configurations; and configuring the honeypot according to the selected configuration," as recited in claim 15.
However, examiner respectfully disagrees with the applicant because McGrew teaches that training the LLM by leveraging malware data stored in a storage, where the LLM references the malware data to identify potential threats in a network, The network device also includes one or more processors configured to execute the computer-readable instructions to generate a plurality of deceptive information using an LLM, configured to attract and engage potential attackers, where the plurality of deceptive information includes one or more characteristics referencing vulnerabilities of the network, continuously monitor for interactions initiated by an interacting party with one or more components of the generated deceptive information, where the interaction is identified as a potential threat to the network, in response to detection of an interaction identified as a potential threat, extract interaction data associated with the interacting party retrieved during the interaction, and retrain the LLM with the interaction data to create more effective honeypots, With the extraction of the interaction dataset, the LLM within the threat management service 102 undergoes a retraining process. The LLM can utilize this interaction data to adapt its language generation capabilities and behavioral analysis. During retraining, the LLM can learn to recognize and understand new attack patterns and tactics exhibited by the potential attacker, After undergoing this retraining, the LLM becomes more skilled in creating deceptive content to interact with potential attackers. As a result, the LLM can more accurately imitate the behavior of attackers, generating responses that are persuasive and genuine. This makes it more difficult for attackers to distinguish between legitimate users and deceptive elements Therefore, McGrew’s system meets the claim limitation.
The applicant argues in page 6 of remarks, that Lu does not disclose "changing a behavior of the honeypot to a state in which the honeypot was when the large language model no longer made progress on the attack," as recited in claim 12. However, examiner respectfully disagrees with the applicant because Lu teaches that obtaining the preset detection behaviour of the target large language model triggered by the access operation of the target client in the target time window, collecting the behaviour characteristic parameter corresponding to each triggered preset detection behaviour; obtaining all target interaction data between the target client and the target large language model in the target time window such that the access frequency protection is the protection action triggered when the access frequency of the target large language model at the same client end exceeds the set access frequency threshold; the client environment detection is the special detection action triggered when the response page of the target large language model is executed with the preset action; the specific detection action is used for determining whether the present client is a browser according to the environment parameters; the honey pot detection is the counting action of the preset HTML link times in the response page of the client access target large language model; The preset HTML link is displayed in the response page in a manner invisible to the biological user. Therefore, Lu’s system meets the claim limitation.
The applicant argues in page 6 of remarks, that Kawasaki fails to disclose "wherein the conducting the at least one attack on the honeypot includes generating, by the large language model, at least one input for a command line interface that the honeypot simulates, and feeding the at least one generated input into the simulated command line interface," as recited in claim 13. However, examiner respectfully disagrees with the applicant because Kawasaki processor to perform a controlling process to process user input (such as Internet Protocol (IP) address ranges), through a command line interface, or through a graphical user interface (GUI). Or, to process input provided by an external program or script through a command line interface or through an application programming interface (API). The initiation of the process may be triggered through any number of causes including external triggering systems including but not limited to scheduled execution (based on calendar schedule or periodic schedule—e.g., hourly) and event triggering (such as through detection by external network inventory systems, IP address management systems, or authentication systems of new devices added to a network or changes to systems discovered by vulnerability management systems or host-based Intrusion Detection Systems (IDSs)). The process may also be initiated from Virtualization Systems or their controllers, when new non-clone VMs are added to the network) and fingerprint manipulation/conversion engine that receives the results of the discovery engine and processes and converts the fingerprints into forms suitable for use by the honeypot system. This step may include binary or textual conversions, and/or heuristic or filtering approaches to translate fingerprints from discovery engine, thereby enabling the system to utilize engines of different types that may generate fingerprints in different formats, and then convert the fingerprints into a single standard format consistent with the needs of the honeypot system. Therefore, Kawasaki’s system meets the claim limitation.
Regarding dependent claims arguments, said arguments are moot because the applied references are not considered to have alleged differences, and therefore are considered to properly show that for which they were cited.
Conclusion
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to AYUB A MAYE whose telephone number is (571)270-5037. The examiner can normally be reached Monday-Friday 9AM-5PM.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, SHEWAYE GELAGAY can be reached at 571-272-4219. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/AYUB A MAYE/Examiner, Art Unit 2436
/TRONG H NGUYEN/Primary Examiner, Art Unit 2436
