Paper Citation Record · LEDGER

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B

As of 9 August 2026, this Paper Citation Record lists 29 of 29 outbound references and 0 inbound Pith citation observations for arXiv:2607.22545.

A citation records a reference. It does not transfer a finding from one paper to another.

pith.paper-citation-record.v1
2607.22545 v1

Coverage vector

measured 29 of 29 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-08-02T14:52:12.628404Z

measured 29 of 29 standing notices

One-hop event checks from named stored sources.

Source: scholarly_work_events, retraction_status_cache, observed 2026-08-09T06:31:02.800959+00:00

measured 0 of 0 inbound itemization

Pith citing papers itemized under the disclosed page cap.

Source: paper_references, paper_reference_links

measured 0 of 1 external citation measurements

A source-named dated measurement, never combined with another source.

Source: cited_works

Reference resolution

29 of 29 outbound references displayed

verified exact0
verified fuzzy0
unresolved29
parse uncertain0
malformed identifier0
metadata mismatch0

External citation measurements

No source-named external measurement is stored.

Outbound references

Observation 83d69bbd-be44-4869-9f7c-db4af95b53a5 · outbound

This paper cites DeBERTaV3: Improving DeBERTa using ELECTRA-Style Pre-Training with Gradient-Disentangled Embedding Sharing.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B DeBERTaV3: Improving DeBERTa using ELECTRA-Style Pre-Training with Gradient-Disentangled Embedding Sharing

Reference 1

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.512497Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 0a51ac02-aab5-4a6c-83a5-51bfa31645ae · outbound

This paper cites Llama Guard: LLM-based Input-Output Safeguard for Human-AI Conversations.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B Llama Guard: LLM-based Input-Output Safeguard for Human-AI Conversations

Reference 2

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.517621Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation bb59b46b-e2a1-46ba-9169-af54eb0c82bc · outbound

This paper cites Llama Guard 3 8B Model Card,.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B Llama Guard 3 8B Model Card,

Reference 3

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.522550Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b28a13af-8bc0-4450-9608-587542258f0c · outbound

This paper cites Granite Guardian.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B Granite Guardian

Reference 4

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.527253Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5421efe7-1cea-4249-bf1c-197444273e0c · outbound

This paper cites WildGuard: Open One-Stop Moderation Tools for Safety Risks, Jailbreaks, and Refusals of LLMs.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B WildGuard: Open One-Stop Moderation Tools for Safety Risks, Jailbreaks, and Refusals of LLMs

Reference 5

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.532422Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e1b881cd-e159-4812-9fe2-5a518ad0890d · outbound

This paper cites PromptGuard 2: Robust Prompt Injection and Jailbreak Detection,.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B PromptGuard 2: Robust Prompt Injection and Jailbreak Detection,

Reference 6

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.536512Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation bcefd773-1b77-425e-8200-66404ab29744 · outbound

This paper cites AprielGuard: A Lightweight Safety Classifier for Enterprise AI,.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B AprielGuard: A Lightweight Safety Classifier for Enterprise AI,

Reference 7

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.540363Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8de2d996-0ef0-4a84-929d-789e0909af53 · outbound

This paper cites OWASP Top 10 for Large Language Model Applications,.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B OWASP Top 10 for Large Language Model Applications,

Reference 8

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.545108Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 1ef3c163-70bd-4ad0-b7ac-f9b0bae8b292 · outbound

This paper cites Domain Shift Amplifies False Positive Rates in LLM Safety Classifiers: Evidence from Financial-Services Deployments,.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B Domain Shift Amplifies False Positive Rates in LLM Safety Classifiers: Evidence from Financial-Services Deployments,

Reference 9

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.549268Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8ed5b653-f27b-4773-9f23-f0da7c00fae5 · outbound

This paper cites PINT: Prompt Injection Test,.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B PINT: Prompt Injection Test,

Reference 10

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.553455Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b96e1c78-9060-44b1-8d8a-4f50130564ce · outbound

This paper cites Ignore Previous Prompt: Attack Techniques For Language Models.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B Ignore Previous Prompt: Attack Techniques For Language Models

Reference 11

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.557207Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e6dc450b-2587-40ad-8e94-f381050cfa06 · outbound

This paper cites Gandalf: Prompt Extraction Challenge,.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B Gandalf: Prompt Extraction Challenge,

Reference 12

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.562062Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b699f868-d796-46df-add4-9f195bfc5f19 · outbound

This paper cites Mosscap: Escalating-Defense Prompt-Injection Benchmark,.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B Mosscap: Escalating-Defense Prompt-Injection Benchmark,

Reference 13

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.566222Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 0c1ab8a3-1312-4bf6-a06d-24d07b79121e · outbound

This paper cites Universal and Transferable Adversarial Attacks on Aligned Language Models.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B Universal and Transferable Adversarial Attacks on Aligned Language Models

Reference 14

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.570231Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d84c92da-2222-42a2-8752-5f263b3c9b38 · outbound

This paper cites AART: AI-Assisted Red-Teaming with Diverse Data Generation for New LLM-powered Applications.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B AART: AI-Assisted Red-Teaming with Diverse Data Generation for New LLM-powered Applications

Reference 15

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.574580Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation f2fbe15c-2b55-4ad7-b185-e424e927d304 · outbound

This paper cites LLM-Mediated Domain-Specific Voice Agents: The Case of TextileBot.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B LLM-Mediated Domain-Specific Voice Agents: The Case of TextileBot

Reference 16

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.578738Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d3c15edd-cf71-4f14-ac06-637c4675ff21 · outbound

This paper cites HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal

Reference 17

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.582991Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 3f1a0b64-2096-4c6d-8cce-6397fee83a69 · outbound

This paper cites Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!

Reference 18

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.587112Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 048e4edc-5e2d-46cd-aa71-38c0bc50b6ab · outbound

This paper cites ToxicChat: Unveiling Hidden Challenges of Toxicity Detection in Real-World User-AI Conversation.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B ToxicChat: Unveiling Hidden Challenges of Toxicity Detection in Real-World User-AI Conversation

Reference 19

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.591220Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7f123c3d-e33c-4331-9ff0-0f7a25b04ebe · outbound

This paper cites BeaverTails: Towards Improved Safety Alignment of LLM via a Human-Preference Dataset,.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B BeaverTails: Towards Improved Safety Alignment of LLM via a Human-Preference Dataset,

Reference 20

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.594827Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 1bd7fe88-6b44-44b1-9e7c-c53df2ebe41c · outbound

This paper cites SALAD-Bench: A Hierarchical and Comprehensive Safety Benchmark for Large Language Models.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B SALAD-Bench: A Hierarchical and Comprehensive Safety Benchmark for Large Language Models

Reference 21

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.598561Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 9f208a3a-3e98-4988-bd5a-a9a9a43da872 · outbound

This paper cites SimpleSafetyTests: a Test Suite for Identifying Critical Safety Risks in Large Language Models.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B SimpleSafetyTests: a Test Suite for Identifying Critical Safety Risks in Large Language Models

Reference 22

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.602287Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation af7225e6-0900-4190-b5cf-68831d5081bd · outbound

This paper cites AgentHarm: A Benchmark for Measuring Harmfulness of LLM Agents.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B AgentHarm: A Benchmark for Measuring Harmfulness of LLM Agents

Reference 23

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.605850Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation f1520880-211c-44f5-b112-f2f0bff72d37 · outbound

This paper cites Probable Inference, the Law of Succession, and Statistical Inference,.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B Probable Inference, the Law of Succession, and Statistical Inference,

Reference 24

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.609376Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5ae4f87f-55bd-4dba-8461-76c2976be57e · outbound

This paper cites The WMDP Benchmark: Measuring and Reducing Malicious Use With Unlearning.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B The WMDP Benchmark: Measuring and Reducing Malicious Use With Unlearning

Reference 25

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.612861Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5266b85b-aef4-49b9-8186-17bb32c29e55 · outbound

This paper cites Investment Adviser Code of Ethics,.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B Investment Adviser Code of Ethics,

Reference 26

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.616537Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 158dd0a9-26ca-4285-8e8d-ac31c00b65bc · outbound

This paper cites Conduct of Business Sourcebook (COBS),.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B Conduct of Business Sourcebook (COBS),

Reference 27

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.620450Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 87f50105-aa15-4d34-9fe0-bcf6efe72e2b · outbound

This paper cites Markets in Financial Instruments Directive II (MiFID II),.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B Markets in Financial Instruments Directive II (MiFID II),

Reference 28

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.623954Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation f5c94133-41b8-4610-bb03-19e00d8f2493 · outbound

This paper cites Artificial Intelligence Act,.

Semalith v1.4: A Calibrated 184M Safety Classifier Achieving State-of-the-Art Prompt-Injection Detection at 44x Fewer Parameters than Llama-Guard-3-8B Artificial Intelligence Act,

Reference 29

Resolution: unresolved; no resolver link, observed 2026-08-02T14:52:12.628404Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Pith citing papers

No inbound Pith citation observations are available.