DETAILED ACTION
The following claims are pending in this office action: 1-20
The following claims are amended: 1, 8 and 15
The following claims are new: -
The following claims are cancelled: -
Claims 1-20 are rejected. This rejection is FINAL.
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 .
RESPONSE TO ARGUMENTS
Applicant’s arguments in the amendment filed 11/12/2025 have been fully considered but are moot in view of new grounds of rejection necessitated by amendment.
Applicant notes: that the cited references do not teach or suggest the amended limitation of claim 1. This limitation is disclosed by Pianta et al. (US Pub. 2025/0182097) as explained below and rejected accordingly.
Independent claims 8 and 15 are amended in a similar way to claim 1. The amended limitations are disclosed by Pianta et al. (US Pub. 2025/0182097) as explained below and rejected accordingly.
Dependent claims 2-7, 9-14 and 16-20 depend on independent claims 1, 8 and 15. The amended elements in the claims are disclosed by Pianta et al. (US Pub. 2025/0182097) as explained below, and so any additional features to the dependent claims are rejected accordingly.
Claim Rejections - 35 USC § 103
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.
Claims 1, 8 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Rideout et al. (US Patent No. 12,248,883) (hereinafter “Rideout”) in view of Salem et al. (US Pub. 2025/0175497) (hereinafter “Salem”) in view of Joseph Durairaj et al. (US Pub. 2017/0032130) (hereinafter “Durairaj”) and in view of Pianta et al. (US Pub. 2025/0182097) (hereinafter “Pianta”)
As per claim 1, Rideout teaches a computing platform, comprising: ([Rideout, col. 9, ln. 58 to col. 10, ln. 5] “a system ... a security platform [computing platform] for machine learning model architectures having a model environment 140 including a local analysis engine 152 .... analysis engines ... instantiate a prompt injection classifier”)
at least one processor; ([Rideout, col. 3, ln. 15-16] “computer systems ... described that may include one or more data processors”)
a communication interface communicatively coupled to the at least one processor; and ([Rideout, col. 3, ln. 16-26] “data processors ... coupled ... connected and can exchange data and/or commands or other instructions or the like [communicatively coupled] via one or more connections [communications interface]”)
memory storing computer-readable instructions that, when executed by the at least one processor, cause the computing platform to: ([Rideout, col. 3, ln. 17-20] “memory ... store instructions that cause at least one processor to perform one or more of the operations described herein”)
receive a prompt injection request; ([Rideout, col. 10, ln. 7-19] “data characterizing a prompt ... is received ...This data can comprise the prompt itself ... the prompt comprises ... a prompt injection”)
segment the prompt injection request; ([Rideout, col. 9, ln. 49-54] “the analysis engine ... preprocess [segment] incoming prompts [the prompt injection request] so that they are suitable for ingestion by the prompt injection classifier ... the raw/original prompt is transformed into sentence embeddings and then input”)
determine if the prompt injection request is an unknown prompt injection request; and ([Rideout, col. 10, ln. 31-33] “the analysis engine 170 can provide the determination to the remediation engine 180”; [col. 2, ln. 11-24] “the prompt injection classifier can be a multi-class model ... With such an arrangement, an analysis engine receives data characterizing a prompt ... The analysis engine uses the prompt injection classifier to determine a category for the prompt which is indicative of whether the prompt comprises or elicits malicious content ... the category can specify a threat severity ... unknown [an unknown prompt injection request]”)
if the prompt injection request is determined to be an unknown prompt injection request, determine if learnings are required; ([Rideout, col. 10, ln. 31-33] “the analysis engine 170 ... provide the determination to the remediation engine 180; [col. 5, ln. 18-22] “remediation engine 180 ... remediation actions in response to a determination of a query as being malicious [determined to be an unknown prompt injection request as per above]”; [col. 5, ln. 25-33] “, the remediation engine 180 ... cause ... the output ... to further analysis [determine if more learnings are required]”; determine if learnings are required for execution of the received prompt injection request is more clearly taught by Salem below)
Rideout does not clearly teach determine if learnings are required for execution of the received prompt injection request; if new learnings are required for execution of the prompt injection request, generate knowledge graphs; and determine at least one new rule based on the generated knowledge graphs, the determined new rule for preventing prompt injection attacks associated with prompt injection requests.
However, Salem teaches determine if learnings are required for execution of the received prompt injection request; ([Salem, para. 0042] “the prompt variation evaluation model 130 determines how effective or successful each variant prompt injection attack was [determines successfulness] ... the prompt variation evaluation model 130 generates prompt variation effectiveness scores 132 [determines successfulness score] for the variant prompt injection attacks [for execution of the received prompt injection request]”; [para. 0085] “instructions 422 include a set of guidelines for improving previous variants, such as learning from previous high-scoring variants [determines if learnings are required if successfulness score is high]”)
if new learnings are required for execution of the prompt injection request, generate knowledge; and ([Salem, para. 0043] “If the prompt variation effectiveness scores 132 satisfy or meet an effectiveness threshold, the attack defense system may utilize the variant prompt injection attacks 112 to improve the robustness [generate knowledge] of the targeted LGM 120 ... details for improving the defense robustness of the targeted LGM based on the set of variant prompt injection attacks are provided in connection with FIG. 7”; [para. 0148] “the robustness measures 712 include model fine-tuning ... generates a training dataset [knowledge]”; if new learnings are required, generate knowledge graphs is more clearly taught by Durairaj below)
determine at least one new rule based on the generated knowledge ... the determined new rule for preventing prompt injection attacks associated with prompt injection requests. ([Salem, para. 0149] “the defense robust model 710 provides robustness measures 712 [based on the generated knowledge] to the targeted LGM 240 that implement guardrail updates [determine at least one new rule] to more accurately detect prompt injection attacks and variants [preventing prompt injection attacks] that are new to the targeted LGM 240 [associated with prompt injection requests – see para. 0030: targeted LGM refers to an LGM that is provided with a prompt that includes an injection attack]”; determine at least one new rule based on the generated knowledge graphs is more clearly taught by Durairaj below)
It would have been obvious before the effective filing date of the claimed invention for one of ordinary skill in the art to have modified the elements disclosed by Rideout with the teachings of Salem to include determine if learnings are required for execution of the received prompt injection request; if new learnings are required for execution of the prompt injection request, generate knowledge; and determine at least one new rule based on the generated knowledge ... the determined new rule for preventing prompt injection attacks associated with prompt injection requests. One of ordinary skill in the art would have been motivated to make this modification because such a technique would provide the benefit improving computing security and accuracy by preventing threat actors from improperly manipulating a targeted LGM to generate unapproved output by allowing the targeted LGM to detect attack variants. (Salem, para. 0016)
Rideout in view of Salem does not clearly teach if new learnings are required, generate knowledge graphs; and determine at least one new rule based on the generated knowledge graphs.
However, Durairaj teaches if new learnings are required, ([Durairaj, para. 0086] “at block 902 ... using trained classifiers [learning] to detect an anomaly [determining new learning] ... The classifiers ... trained to learn patterns of clusters based on training event”) generate knowledge graphs; and ([para. 0087] “At block 904 ... include generating a predictive attack graph based on the detected anomaly [if new learnings are required]”)
determine at least one new rule based on the generated knowledge graphs ([Durairaj, para. 0096] “the method 900 may further include determining a path in the predictive attack graph ... determining an occurrence of an attack associated with the path ... and creating an ephemeral rule [determine at least one new rule] based on the occurrence of the attack [based on the generated knowledge graphs]”)
It would have been obvious before the effective filing date of the claimed invention for one of ordinary skill in the art to have modified the elements disclosed by Rideout in view of Salem with the teachings of Durairaj to include if new learnings are required, generate knowledge graphs; and determine at least one new rule based on the generated knowledge graphs. One of ordinary skill in the art would have been motivated to make this modification because such a technique would provide the benefit of adding visibility to event patterns, and to correlate sentiment analysis with anomalies in order to minimize the net value of compromised assets. (Durairaj, para. 0023)
Rideout in view of Salem and Durairaj does not clearly teach analyze the at least one new rule to determine an impact on an enterprise based on key performance metrics or organizational health factors before approval and implementation of the at least one new rule.
However, Pianta teaches analyze the at least one new rule to determine an impact on an enterprise based on key performance metrics or organizational health factors ([Pianta, para. 0091] “performance analyzer 416 of rule evaluator 406 may apply a heuristic rule generated by rule generator 402 to a set of transactions to determine [analyze the at least one new rule] a set of block transaction and a set of unblocked transactions [key performance metrics] to determine whether performance of the rule is acceptable [an impact on an enterprise – see para. 0062: “predict the particular outcome ... using the rule ... until an acceptable area ... is achieved for the ... group of merchants”] ... Data based on the set of blocked transactions and the set of unblocked transactions include various metrics, such as ... block rate and false positive rate [key performance metrics]”) before approval and implementation of the at least one new rule. ([Para. 0085] “once a user is satisfied with the performance ... the users may provide input that causes the ... rule to be implemented”)
It would have been obvious before the effective filing date of the claimed invention for one of ordinary skill in the art to have modified the elements disclosed by Rideout in view of Salem and Durairaj with the teachings of Pianta to include analyze the at least one new rule to determine an impact on an enterprise based on key performance metrics or organizational health factors before approval and implementation of the at least one new rule. One of ordinary skill in the art would have been motivated to make this modification because such a technique would provide the benefit of increase the accessibility, practicality, adaptability, and availability of processing rules, thereby improving the functioning of a server systems for generating processing rules as compared to conventional approaches. (Durairaj, para. 0023)
As per claim 8, Rideout teaches a method, comprising: ([Rideout, col. 12, ln. 7-8] “The subject matter described herein can be embodied in ... methods”)
at a computing platform comprising at least one processor, a communication interface, and memory: ([Rideout, col. 9, ln. 58 to col. 10, ln. 5] “a system ... a security platform [computing platform] for machine learning model architectures having a model environment 140 including a local analysis engine 152 .... analysis engines ... instantiate a prompt injection classifier”; [col. 3, ln. 15-16] “computer systems ... described that may include one or more data processors”; [col. 3, ln. 16-26] “data processors ... coupled ... connected and can exchange data and/or commands or other instructions or the like [communicatively coupled] via one or more connections [communications interface]”; [col. 3, ln. 17-20] “memory ... store instructions that cause at least one processor to perform one or more of the operations described herein”)
receiving a prompt injection request; ([Rideout, col. 10, ln. 7-19] “data characterizing a prompt ... is received ...This data can comprise the prompt itself ... the prompt comprises ... a prompt injection”)
segmenting the prompt injection request; ([Rideout, col. 9, ln. 49-54] “the analysis engine ... preprocess [segmenting] incoming prompts [the prompt injection request] so that they are suitable for ingestion by the prompt injection classifier ... the raw/original prompt is transformed into sentence embeddings and then input”)
determining if (examiner interprets the following “if” limitations as contingent limitations; as such, the broadest reasonable interpretation of a method claim having contingent limitations requires only those steps that must be performed; here, Rideout also discloses the prompt injection is determined not to be “an unknown prompt injection request”; [Rideout, col. 9, ln. 10-15] “a binary classifier can predict ... a score closer to 0 is indicating that the prompt is benign”; however, examiner interprets that “determining at least one new rule based on the generated knowledge graphs” must be performed as it is not nested within the if statements; thus determining “the prompt injection request is an unknown prompt injection request”, and determining “new learnings are required” are also necessarily performed as both determinations must be made in order to generate the knowledge graphs) the prompt injection request is an unknown prompt injection request; and ([Rideout, col. 10, ln. 31-33] “the analysis engine 170 can provide the determination to the remediation engine 180”; [col. 2, ln. 11-24] “the prompt injection classifier can be a multi-class model ... With such an arrangement, an analysis engine receives data characterizing a prompt ... The analysis engine uses the prompt injection classifier to determine a category for the prompt which is indicative of whether the prompt comprises or elicits malicious content ... the category can specify a threat severity ... unknown [an unknown prompt injection request]”)
if the prompt injection request is determined to be an unknown prompt injection request, determining if learnings are required; ([Rideout, col. 10, ln. 31-33] “the analysis engine 170 ... provide the determination to the remediation engine 180; [col. 5, ln. 18-22] “remediation engine 180 ... remediation actions in response to a determination of a query as being malicious [determined to be an unknown prompt injection request as per above]”; [col. 5, ln. 25-33] “, the remediation engine 180 ... cause ... the output ... to further analysis [determine if more learnings are required]”; determining if learnings are required for execution of the received prompt injection request is more clearly taught by Salem below)
Rideout does not clearly teach determining if learnings are required for execution of the received prompt injection request; if new learnings are required for execution of the prompt injection request, generating knowledge graphs; and determining at least one new rule based on the generated knowledge graphs, the determined new rule for preventing prompt injection attacks associated with prompt injection requests.
However, Salem teaches determining if learnings are required for execution of the received prompt injection request; ([Salem, para. 0042] “the prompt variation evaluation model 130 determines how effective or successful each variant prompt injection attack was [determines successfulness] ... the prompt variation evaluation model 130 generates prompt variation effectiveness scores 132 [determines successfulness score] for the variant prompt injection attacks [for execution of the received prompt injection request]”; [para. 0085] “instructions 422 include a set of guidelines for improving previous variants, such as learning from previous high-scoring variants [determines if learnings are required if successfulness score is high]”)
if new learnings are required for execution of the prompt injection request, generating knowledge; and ([Salem, para. 0043] “If the prompt variation effectiveness scores 132 satisfy or meet an effectiveness threshold, the attack defense system may utilize the variant prompt injection attacks 112 to improve the robustness [generate knowledge] of the targeted LGM 120 ... details for improving the defense robustness of the targeted LGM based on the set of variant prompt injection attacks are provided in connection with FIG. 7”; [para. 0148] “the robustness measures 712 include model fine-tuning ... generates a training dataset [knowledge]”; if new learnings are required, generating knowledge graphs is more clearly taught by Durairaj below)
determining at least one new rule based on the generated knowledge ... the determined new rule for preventing prompt injection attacks associated with prompt injection requests. ([Salem, para. 0149] “the defense robust model 710 provides robustness measures 712 [based on the generated knowledge] to the targeted LGM 240 that implement guardrail updates [determine at least one new rule] to more accurately detect prompt injection attacks and variants [preventing prompt injection attacks] that are new to the targeted LGM 240 [associated with prompt injection requests – see para. 0030: targeted LGM refers to an LGM that is provided with a prompt that includes an injection attack]”; determining at least one new rule based on the generated knowledge graphs is more clearly taught by Durairaj below)
It would have been obvious before the effective filing date of the claimed invention for one of ordinary skill in the art to have modified the elements disclosed by Rideout with the teachings of Salem to include determining if learnings are required for execution of the received prompt injection request; if new learnings are required for execution of the prompt injection request, generating knowledge; and determining at least one new rule based on the generated knowledge ... the determined new rule for preventing prompt injection attacks associated with prompt injection requests. One of ordinary skill in the art would have been motivated to make this modification because such a technique would provide the benefit improving computing security and accuracy by preventing threat actors from improperly manipulating a targeted LGM to generate unapproved output by allowing the targeted LGM to detect attack variants. (Salem, para. 0016)
Rideout in view of Salem does not clearly teach if new learnings are required, generating knowledge graphs; and determining at least one new rule based on the generated knowledge graphs.
However, Durairaj teaches if new learnings are required, ([Durairaj, para. 0086] “at block 902 ... using trained classifiers [learning] to detect an anomaly [determining new learning] ... The classifiers ... trained to learn patterns of clusters based on training event”) generating knowledge graphs; and ([para. 0087] “At block 904 ... include generating a predictive attack graph based on the detected anomaly [if new learnings are required]”)
determining at least one new rule based on the generated knowledge graphs. ([Durairaj, para. 0096] “the method 900 may further include determining a path in the predictive attack graph ... determining an occurrence of an attack associated with the path ... and creating an ephemeral rule [determine at least one new rule] based on the occurrence of the attack [based on the generated knowledge graphs]”)
It would have been obvious before the effective filing date of the claimed invention for one of ordinary skill in the art to have modified the elements disclosed by Rideout in view of Salem with the teachings of Durairaj to include if new learnings are required, generating knowledge graphs; and determining at least one new rule based on the generated knowledge graphs. One of ordinary skill in the art would have been motivated to make this modification because such a technique would provide the benefit of adding visibility to event patterns, and to correlate sentiment analysis with anomalies in order to minimize the net value of compromised assets. (Durairaj, para. 0023)
Rideout in view of Salem and Durairaj does not clearly teach analyze the at least one new rule to determine an impact on an enterprise based on key performance metrics or organizational health factors before approval and implementation of the at least one new rule.
However, Pianta teaches analyze the at least one new rule to determine an impact on an enterprise based on key performance metrics or organizational health factors ([Pianta, para. 0091] “performance analyzer 416 of rule evaluator 406 may apply a heuristic rule generated by rule generator 402 to a set of transactions to determine [analyze the at least one new rule] a set of block transaction and a set of unblocked transactions [key performance metrics] to determine whether performance of the rule is acceptable [an impact on an enterprise – see para. 0062: “predict the particular outcome ... using the rule ... until an acceptable area ... is achieved for the ... group of merchants”] ... Data based on the set of blocked transactions and the set of unblocked transactions include various metrics, such as ... block rate and false positive rate [key performance metrics]”) before approval and implementation of the at least one new rule. ([Para. 0085] “once a user is satisfied with the performance ... the users may provide input that causes the ... rule to be implemented”)
It would have been obvious before the effective filing date of the claimed invention for one of ordinary skill in the art to have modified the elements disclosed by Rideout in view of Salem and Durairaj with the teachings of Pianta to include analyze the at least one new rule to determine an impact on an enterprise based on key performance metrics or organizational health factors before approval and implementation of the at least one new rule. One of ordinary skill in the art would have been motivated to make this modification because such a technique would provide the benefit of increase the accessibility, practicality, adaptability, and availability of processing rules, thereby improving the functioning of a server systems for generating processing rules as compared to conventional approaches. (Durairaj, para. 0023)
As per claim 15, Rideout teaches one or more non-transitory computer-readable media storing instructions that, when executed by a computing platform comprising at least one processor, a communication interface, and memory, cause the computing platform to: ([Rideout, col. 3, ln. 10-15] “Non-transitory computer program products (i.e., physically embodied computer program products) are also described that comprise instructions, which when executed by one or more data processors of one or more computing systems, cause at least one data processor to perform operations herein”)
receive a prompt injection request; ([Rideout, col. 10, ln. 7-19] “data characterizing a prompt ... is received ...This data can comprise the prompt itself ... the prompt comprises ... a prompt injection”)
segment the prompt injection request; ([Rideout, col. 9, ln. 49-54] “the analysis engine ... preprocess [segment] incoming prompts [the prompt injection request] so that they are suitable for ingestion by the prompt injection classifier ... the raw/original prompt is transformed into sentence embeddings and then input”)
determine if the prompt injection request is an unknown prompt injection request; and ([Rideout, col. 10, ln. 31-33] “the analysis engine 170 can provide the determination to the remediation engine 180”; [col. 2, ln. 11-24] “the prompt injection classifier can be a multi-class model ... With such an arrangement, an analysis engine receives data characterizing a prompt ... The analysis engine uses the prompt injection classifier to determine a category for the prompt which is indicative of whether the prompt comprises or elicits malicious content ... the category can specify a threat severity ... unknown [an unknown prompt injection request]”)
if the prompt injection request is determined to be an unknown prompt injection request, determine if learnings are required; ([Rideout, col. 10, ln. 31-33] “the analysis engine 170 ... provide the determination to the remediation engine 180; [col. 5, ln. 18-22] “remediation engine 180 ... remediation actions in response to a determination of a query as being malicious [determined to be an unknown prompt injection request as per above]”; [col. 5, ln. 25-33] “, the remediation engine 180 ... cause ... the output ... to further analysis [determine if more learnings are required]”; determine if learnings are required for execution of the received prompt injection request is more clearly taught by Salem below)
Rideout does not clearly teach determine if learnings are required for execution of the received prompt injection request; if new learnings are required for execution of the prompt injection request, generate knowledge graphs; and determine at least one new rule based on the generated knowledge graphs, the determined new rule for preventing prompt injection attacks associated with prompt injection requests.
However, Salem teaches determine if learnings are required for execution of the received prompt injection request; ([Salem, para. 0042] “the prompt variation evaluation model 130 determines how effective or successful each variant prompt injection attack was [determines successfulness] ... the prompt variation evaluation model 130 generates prompt variation effectiveness scores 132 [determines successfulness score] for the variant prompt injection attacks [for execution of the received prompt injection request]”; [para. 0085] “instructions 422 include a set of guidelines for improving previous variants, such as learning from previous high-scoring variants [determines if learnings are required if successfulness score is high]”)
if new learnings are required for execution of the prompt injection request, generate knowledge; and ([Salem, para. 0043] “If the prompt variation effectiveness scores 132 satisfy or meet an effectiveness threshold, the attack defense system may utilize the variant prompt injection attacks 112 to improve the robustness [generate knowledge] of the targeted LGM 120 ... details for improving the defense robustness of the targeted LGM based on the set of variant prompt injection attacks are provided in connection with FIG. 7”; [para. 0148] “the robustness measures 712 include model fine-tuning ... generates a training dataset [knowledge]”; if new learnings are required, generate knowledge graphs is more clearly taught by Durairaj below)
determine at least one new rule based on the generated knowledge ... the determined new rule for preventing prompt injection attacks associated with prompt injection requests. ([Salem, para. 0149] “the defense robust model 710 provides robustness measures 712 [based on the generated knowledge] to the targeted LGM 240 that implement guardrail updates [determine at least one new rule] to more accurately detect prompt injection attacks and variants [preventing prompt injection attacks] that are new to the targeted LGM 240 [associated with prompt injection requests – see para. 0030: targeted LGM refers to an LGM that is provided with a prompt that includes an injection attack]”; determine at least one new rule based on the generated knowledge graphs is more clearly taught by Durairaj below)
It would have been obvious before the effective filing date of the claimed invention for one of ordinary skill in the art to have modified the elements disclosed by Rideout with the teachings of Salem to include determine if learnings are required for execution of the received prompt injection request; if new learnings are required for execution of the prompt injection request, generate knowledge; and determine at least one new rule based on the generated knowledge ... the determined new rule for preventing prompt injection attacks associated with prompt injection requests. One of ordinary skill in the art would have been motivated to make this modification because such a technique would provide the benefit improving computing security and accuracy by preventing threat actors from improperly manipulating a targeted LGM to generate unapproved output by allowing the targeted LGM to detect attack variants. (Salem, para. 0016)
Rideout in view of Salem does not clearly teach if new learnings are required, generate knowledge graphs; and determine at least one new rule based on the generated knowledge graphs.
However, Durairaj teaches if new learnings are required, ([Durairaj, para. 0086] “at block 902 ... using trained classifiers [learning] to detect an anomaly [determining new learning] ... The classifiers ... trained to learn patterns of clusters based on training event”) generate knowledge graphs; and ([para. 0087] “At block 904 ... include generating a predictive attack graph based on the detected anomaly [if new learnings are required]”)
determine at least one new rule based on the generated knowledge graphs. ([Durairaj, para. 0096] “the method 900 may further include determining a path in the predictive attack graph ... determining an occurrence of an attack associated with the path ... and creating an ephemeral rule [determine at least one new rule] based on the occurrence of the attack [based on the generated knowledge graphs]”)
It would have been obvious before the effective filing date of the claimed invention for one of ordinary skill in the art to have modified the elements disclosed by Rideout in view of Salem with the teachings of Durairaj to include if new learnings are required, generate knowledge graphs; and determine at least one new rule based on the generated knowledge graphs. One of ordinary skill in the art would have been motivated to make this modification because such a technique would provide the benefit of adding visibility to event patterns, and to correlate sentiment analysis with anomalies in order to minimize the net value of compromised assets. (Durairaj, para. 0023)
Rideout in view of Salem and Durairaj does not clearly teach analyze the at least one new rule to determine an impact on an enterprise based on key performance metrics or organizational health factors before approval and implementation of the at least one new rule.
However, Pianta teaches analyze the at least one new rule to determine an impact on an enterprise based on key performance metrics or organizational health factors ([Pianta, para. 0091] “performance analyzer 416 of rule evaluator 406 may apply a heuristic rule generated by rule generator 402 to a set of transactions to determine [analyze the at least one new rule] a set of block transaction and a set of unblocked transactions [key performance metrics] to determine whether performance of the rule is acceptable [an impact on an enterprise – see para. 0062: “predict the particular outcome ... using the rule ... until an acceptable area ... is achieved for the ... group of merchants”] ... Data based on the set of blocked transactions and the set of unblocked transactions include various metrics, such as ... block rate and false positive rate [key performance metrics]”) before approval and implementation of the at least one new rule. ([Para. 0085] “once a user is satisfied with the performance ... the users may provide input that causes the ... rule to be implemented”)
It would have been obvious before the effective filing date of the claimed invention for one of ordinary skill in the art to have modified the elements disclosed by Rideout in view of Salem and Durairaj with the teachings of Pianta to include analyze the at least one new rule to determine an impact on an enterprise based on key performance metrics or organizational health factors before approval and implementation of the at least one new rule. One of ordinary skill in the art would have been motivated to make this modification because such a technique would provide the benefit of increase the accessibility, practicality, adaptability, and availability of processing rules, thereby improving the functioning of a server systems for generating processing rules as compared to conventional approaches. (Durairaj, para. 0023)
Claims 2-7, 9-14 and 16-20 are rejected under 35 U.S.C. 103 as being unpatentable over Rideout in view of Salem, Durairaj and Pianta as applied to claims 1, 8 and 15 above and further in view of Horesh et al. (US Pub. 2025/0278630) (hereinafter “Horesh”)
As per claim 2, Rideout in view of Salem, Durairaj and Pianta teaches claim 1.
Rideout in view of Salem, Durairaj and Pianta does not clearly teach wherein the memory stores additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to: score the at least one new rule to determine if the at least one new rule reaches a predetermined threshold for implementing the at least one new rule.
However, Horesh teaches wherein the memory stores additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to: ([Horesh, para. 0030] “memory that stores instructions that, when executed by the at least one processor, cause the processor to perform the functionality of each component when executed”)
score the at least one new rule ([Horesh, para. 0036] “The score component 320 is configured to score prompt candidates [new rule]”; [para. 0003] Prompts are rules as “Prompts serve as guilding instructions to direct an LLM’s comprehension and response generation process”) to determine if the at least one new rule reaches a predetermined threshold for implementing the at least one new rule. ([Para. 0038] “The selection component 330 is configured to select one or more prompts based on a score computed by the score component 320 for each individual prompt ... One or more thresholds [predetermined threshold] can be utilized by the selection component 330 to select the one or more prompts for ... return [implementing] as a final output prompt [the at least one new rule]”)
It would have been obvious before the effective filing date of the claimed invention for one of ordinary skill in the art to have modified the elements disclosed by Rideout in view of Salem, Durairaj and Pianta with the teachings of Horesh to include wherein the memory stores additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to: score the at least one new rule to determine if the at least one new rule reaches a predetermined threshold for implementing the at least one new rule. One of ordinary skill in the art would have been motivated to make this modification because determining an effective rule is an important technical step for generating a desired output from an LLM and an efficient rule (as determined by the score) has the beneficial technical effect of reducing the occurrence of irrelevant and potentially harmful results produced. (Horesh, para. 0014)
As per claim 3, Rideout in view of Salem, Durairaj and Pianta teaches claim 2.
Rideout in view of Salem, Durairaj and Pianta does not clearly teach wherein scoring the at least one new rule comprises scoring the at least new rule based on security factors.
However, Horesh teaches wherein scoring the at least one new rule comprises scoring the at least new rule based on security factors. ([Horesh, para. 0036] “The score component 320 is configured to score prompt candidates using one or more metrics ... metrics can include ... the vulnerability of the prompt to adversarial attack ... e.g., secure or insecure prompts”)
It would have been obvious before the effective filing date of the claimed invention for one of ordinary skill in the art to have modified the elements disclosed by Rideout in view of Salem, Durairaj and Pianta with the teachings of Horesh to include wherein scoring the at least one new rule comprises scoring the at least new rule based on security factors. One of ordinary skill in the art would have been motivated to make this modification because an objective scoring mechanism can be provided to measure prompt performance in terms of security, providing the benefit of reducing or eliminating reliance on subjective opinion. (Horesh, para. 0017)
As per claim 4, Rideout in view of Salem, Durairaj and Pianta teaches claim 3.
Rideout in view of Salem, Durairaj and Pianta does not clearly teach wherein scoring the at least one new rule comprises scoring the at least one new rule based on sustainability factors.
However, Horesh teaches wherein scoring the at least one new rule comprises scoring the at least one new rule based on sustainability factors. ([Horesh, para. 0036] “The score component 320 is configured to score prompt candidates using one or more metrics ... metrics can include ... the vulnerability of the prompt ... ignoring previous instructions or content moderation guidelines”; these are sustainability factors as ignoring such instructions/guidelines would necessarily make current prompts unsustainable/incapable of improvement in view of future prompts)
It would have been obvious before the effective filing date of the claimed invention for one of ordinary skill in the art to have modified the elements disclosed by Rideout in view of Salem, Durairaj and Pianta with the teachings of Horesh to include wherein scoring the at least one new rule comprises scoring the at least new rule based on sustainability factors. One of ordinary skill in the art would have been motivated to make this modification because objective scoring mechanism can be provided to measure prompt performance in terms of accuracy, providing the benefit of reducing or eliminating reliance on subjective opinion. (Horesh, para. 0017)
As per claim 5, Rideout in view of Salem, Durairaj and Pianta teaches claim 4.
Rideout in view of Salem, Durairaj and Pianta does not clearly teach wherein scoring the at least one new rule comprises scoring the at least one new rule based on revenue factors.
However, Horesh teaches wherein scoring the at least one new rule comprises scoring the at least one new rule based on revenue factors. ([Horesh, para. 0036] “The score component 320 is configured to score prompt candidates using one or more metrics ... metrics can include ... the cost of running the prompt in terms of computational resources required”)
It would have been obvious before the effective filing date of the claimed invention for one of ordinary skill in the art to have modified the elements disclosed by Rideout in view of Salem, Durairaj and Pianta with the teachings of Horesh to include wherein scoring the at least one new rule comprises scoring the at least one new rule based on revenue factors. One of ordinary skill in the art would have been motivated to make this modification because objective scoring mechanism can be provided to measure prompt performance in terms of cost, providing the benefit of reducing or eliminating reliance on subjective opinion. (Horesh, para. 0017)
As per claim 6, Rideout in view of Salem, Durairaj and Pianta teaches claim 5.
Rideout in view of Salem, Durairaj and Pianta does not clearly teach wherein scoring the at least one new rule comprises scoring the at least one new rule based on resilience factors.
However, Horesh teaches wherein scoring the at least one new rule comprises scoring the at least one new rule based on resilience factors. ([Horesh, para. 0036] “The score component 320 is configured to score prompt candidates using one or more metrics ... metrics can include ... and the vulnerability of the prompt to adversarial attack”)
It would have been obvious before the effective filing date of the claimed invention for one of ordinary skill in the art to have modified the elements disclosed by Rideout in view of Salem, Durairaj and Pianta with the teachings of Horesh to include wherein scoring the at least one new rule comprises scoring the at least one new rule based on resilience factors. One of ordinary skill in the art would have been motivated to make this modification because s objective scoring mechanism can be provided to measure prompt performance in terms of safety, providing the benefit of reducing or eliminating reliance on subjective opinion. (Horesh, para. 0017)
As per claim 7, Rideout in view of Salem, Durairaj and Pianta teaches claim 2.
Rideout in view of Salem, Durairaj and Pianta does not clearly teach wherein the memory stores additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to: update a knowledge repository associated with the computer platform based on the at least one new rule meeting or exceeding the predetermined threshold for implementing the at least one new rule.
However, Horesh teaches wherein the memory stores additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to: ([Horesh, para. 0030] “memory that stores instructions that, when executed by the at least one processor, cause the processor to perform the functionality of each component when executed”)
update a knowledge repository associated with the computer platform ([Horesh, para. 0025] “One of the candidate prompts can be selected ... and transmitted to the target machine-learning model 130 [knowledge repository]”; [para. 0057-0058] “processing system 500 configured to perform various aspects described herein ... Processing system 500 is generally an example of an electronic device ... including ... servers [computer platform]”) based on the at least one new rule meeting or exceeding the predetermined threshold for implementing the at least one new rule. ([Horesh, para. 0038] “The selection component 330 is configured to select one or more prompts [update a knowledge repository] based on a score computed by the score component 320 for each individual prompt [new rule] ... One or more thresholds can be utilized by the selection component 330 to select the one or more prompts [implementing the at least one new rule]; [para. 0041] “return a prompt [update a knowledge repository] ... a prompt score [based on the new rule] satisfying [meeting or exceeding] a threshold”)
It would have been obvious before the effective filing date of the claimed invention for one of ordinary skill in the art to have modified the elements disclosed by Rideout in view of Salem, Durairaj and Pianta with the teachings of Horesh to include wherein the memory stores additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to: update a knowledge repository associated with the computer platform based on the at least one new rule meeting or exceeding the predetermined threshold for implementing the at least one new rule. One of ordinary skill in the art would have been motivated to make this modification because such a technique would provide the benefit of iteratively maximizing rule performance which, as a result, allows rule generation to be more efficient, especially when dealing with complex and diverse domains, as a broad range of rule variations can be determined and evaluated quickly. (Horesh, para. 0017)
As per claim 9, the claim language is identical or substantially similar to that of claim 2. Therefore, it is rejected under the same rationale applied to claim 2.
As per claim 10, the claim language is identical or substantially similar to that of claim 3. Therefore, it is rejected under the same rationale applied to claim 3.
As per claim 11, the claim language is identical or substantially similar to that of claim 4. Therefore, it is rejected under the same rationale applied to claim 4.
As per claim 12, the claim language is identical or substantially similar to that of claim 5. Therefore, it is rejected under the same rationale applied to claim 5.
As per claim 13, the claim language is identical or substantially similar to that of claim 6. Therefore, it is rejected under the same rationale applied to claim 6.
As per claim 14, the claim language is identical or substantially similar to that of claim 7. Therefore, it is rejected under the same rationale applied to claim 7.
As per claim 16, the claim language is identical or substantially similar to that of claim 2. Therefore, it is rejected under the same rationale applied to claim 2.
As per claim 17, the claim language is identical or substantially similar to that of claim 3. Therefore, it is rejected under the same rationale applied to claim 3.
As per claim 18, the claim language is identical or substantially similar to that of claim 4. Therefore, it is rejected under the same rationale applied to claim 4.
As per claim 19, the claim language is identical or substantially similar to that of claim 5. Therefore, it is rejected under the same rationale applied to claim 5.
As per claim 20, the claim language is identical or substantially similar to that of claim 6. Therefore, it is rejected under the same rationale applied to claim 6.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
Kanwar et al. (US Pub. 2025/0265460) discloses creating match rules and analyzing match rules and the performance of these rules in a live production environment to add the match rules to the production environment.
McCarthy et al. (US Patent 11,765,207) discloses generating policy statements and determining that the policy statement may be accepted based on various organization requirements
Reagan et al. (US Pub. 2025/0363200) discloses dynamic enforcement of management rules which includes a rules engine that analyzes a subset of provider restrictions and determines restrictions that that apply to the enterprise content.
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). 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 ZHE LIU whose telephone number is (571) 272-3634. The examiner can normally be reached on Monday - Friday: 8:30 AM to 5:30 PM.
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, Carl Colin can be reached on (571) 272-3862. The fax phone number for the organization where this application or proceeding is assigned is (571) 273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see https://ppair-my.uspto.gov/pair/PrivatePair. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at (866) 217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call (800) 786-9199 (IN USA OR CANADA) or (571) 272-1000.
/ZHE LIU/Examiner, Art Unit 2493
/Michael Simitoski/Primary Examiner, Art Unit 2493
Prosecution Insights