ARXIV:2602.15143 · AI SECURITY · SUBMITTED 02 APR · 02:30 UTC · FRESHNESS STALE

VerifiedSource: PDF linkedPartialPaperPack: 3 of 4 citation fields filledMissingMissing fields: authorsPartialProof: unverified proof status

Protecting Language Models Against Unauthorized Distillation through Trace Rewriting

arXiv

A method to modify LLM outputs to prevent unauthorized knowledge distillation and embed verifiable watermarks in student models.

Blocked on Code›Score3.0Evidence unverified

Opportunity summary

Pain A method to modify LLM outputs to prevent unauthorized knowledge distillation and embed verifiable watermarks in student models.

Evidence 0 refs | 0 sources | 17% coverage

Blocker Evidence unverified

Open Build Read PDF Signal Canvas Track

PROBLEM

A method to modify LLM outputs to prevent unauthorized knowledge distillation and embed verifiable watermarks in student models. However, unauthorized use of knowledge distillation takes unfair advantage of the considerable effort and cost put…

METHOD

Full abstract

Knowledge distillation is a widely adopted technique for transferring capabilities from LLMs to smaller, more efficient student models. However, unauthorized use of knowledge distillation takes unfair advantage of the considerable effort and cost put into developing frontier models. We investigate methods for modifying teacher-generated reasoning traces to achieve two objectives that deter unauthorized distillation: (1) \emph{anti-distillation}, or degrading the training usefulness of query responses, and (2) \emph{API watermarking}, which embeds verifiable signatures in student models. We introduce several approaches for dynamically rewriting a teacher's reasoning outputs while preserving answer correctness and semantic coherence. Two of these leverage the rewriting capabilities of LLMs, while others use gradient-based techniques. Our experiments show that a simple instruction-based rewriting approach achieves a strong anti-distillation effect while maintaining or even improving teacher performance. Furthermore, we show that our rewriting approach also enables highly reliable watermark detection with essentially no false alarms.

RESULT

ScienceToStartup currently rates this 3.0/10 on the public viability pass. We investigate methods for modifying teacher-generated reasoning traces to achieve two objectives that deter unauthorized distillation: (1) \emph{anti-distillation}, or degrading the training usefulness of…

WHY NOW

AI Security moved forward this cycle; last verified April 2026. Public score 3.0/10.

Continue into Read for claims, analysis, references, and neighboring papers.

Opportunity summary

Score3.0

PainA method to modify LLM outputs to prevent unauthorized knowledge distillation and embed verifiable watermarks in student models.

Evidence0 refs | 0 sources | 17% coverage

Blockermissing authors

Analysis summary

A method to modify LLM outputs to prevent unauthorized knowledge distillation and embed verifiable watermarks in student models.

VerifiedSource: PDF linkedPartialPaperPack: 3 of 4 citation fields filledMissingMissing fields: authorsPartialProof: unverified proof status

ARXIV:2602.15143 · AI SECURITY · SUBMITTED 02 APR · 02:30 UTC · FRESHNESS STALE

VerifiedSource: PDF linkedPartialPaperPack: 3 of 4 citation fields filledMissingMissing fields: authorsPartialProof: unverified proof status

Protecting Language Models Against Unauthorized Distillation through Trace Rewriting

arXiv

A method to modify LLM outputs to prevent unauthorized knowledge distillation and embed verifiable watermarks in student models.

Blocked on Code›Score3.0Evidence unverified

Opportunity summary

Pain A method to modify LLM outputs to prevent unauthorized knowledge distillation and embed verifiable watermarks in student models.

Evidence 0 refs | 0 sources | 17% coverage

Blocker Evidence unverified

Open Build Read PDF Signal Canvas Track

PROBLEM

METHOD

Full abstract

RESULT

WHY NOW

AI Security moved forward this cycle; last verified April 2026. Public score 3.0/10.

Continue into Read for claims, analysis, references, and neighboring papers.

Opportunity summary

Score3.0

PainA method to modify LLM outputs to prevent unauthorized knowledge distillation and embed verifiable watermarks in student models.

Evidence0 refs | 0 sources | 17% coverage

Blockermissing authors

Analysis summary

A method to modify LLM outputs to prevent unauthorized knowledge distillation and embed verifiable watermarks in student models.

VerifiedSource: PDF linkedPartialPaperPack: 3 of 4 citation fields filledMissingMissing fields: authorsPartialProof: unverified proof status

Paper Pack

10.48550/arXiv.2602.15143

Protecting Language Models Against Unauthorized Distillation through Trace Rewriting

A method to modify LLM outputs to prevent unauthorized knowledge distillation and embed verifiable watermarks in student models.

Abstract

Source availability

PDF linked

The paper record includes a public PDF URL.

Extraction status

Derived fallback

Read summaries are estimated from adjacent metadata, not verified extraction rows.

Proof status

unverified

0 refs; 0 sources; 17% coverage.

What was readable

linkedon filenot materializedderived fallback30 indexednot indexed

Derived fallback: Estimated from adjacent evidence; not verified from source.

Viability

3.0

Time to MVP

MVP estimate missing

Commercial

No commercial flags on file

Export

Preparing verified analysis

lens / founder

PROBLEM

METHOD

RESULT

WHY NOW

AI Security moved forward this cycle; last verified April 2026. Public score 3.0/10.

Claim map

Abstract-backed public claims while anchored extraction refreshes.

Strong 0Mixed 0Weak 4

Evidencepartial
A method to modify LLM outputs to prevent unauthorized knowledge distillation and embed verifiable watermarks in student models. However, unauthorized use of knowledge distillation takes unfair advantage of the considerable effort and cost put into developing frontier models.
Implicationpartial
Abstract-backed fallback claim; anchored extraction has not materialized a public claim row yet.
Verificationpartial
partial
Evidencepartial
Knowledge distillation is a widely adopted technique for transferring capabilities from LLMs to smaller, more efficient student models. However, unauthorized use of knowledge distillation takes unfair advantage of the considerable effort and cost put into developing frontier models.
Implicationpartial
Abstract-backed fallback claim; anchored extraction has not materialized a public claim row yet.
Verificationpartial
partial
Evidencepartial
ScienceToStartup currently rates this 3.0/10 on the public viability pass. We investigate methods for modifying teacher-generated reasoning traces to achieve two objectives that deter unauthorized distillation: (1) \emph{anti-distillation}, or degrading the training usefulness of query responses, and (2) \emph{API watermarking}, which embeds verifiable signatures in student models.
Implicationpartial
Abstract-backed fallback claim; anchored extraction has not materialized a public claim row yet.
Verificationpartial
partial
Evidencepartial
AI Security moved forward this cycle; last verified April 2026. Public score 3.0/10.
Implicationpartial
Abstract-backed fallback claim; anchored extraction has not materialized a public claim row yet.
Verificationpartial
partial

Constellation map

Paper-native neighborhood for concepts, methods, materials, markets, and competitors. Missing lanes stay labeled instead of disappearing behind commercialization gates.

Open full Signal Canvas

Concepts

not indexed

Methods

Materials

PDF linked

Markets

AI Security

Competitors

not indexed

Competitive landscape

A method to modify LLM outputs to prevent unauthorized knowledge distillation and embed verifiable watermarks in student models.

Segment

AI Security

Adoption evidence

No public code link in the paper record yet

Commercial read

3.0/10 public viability

Direct

not classified

Adjacent

not classified

Substitute

not classified

Unknown

not classified

Buzz

No indexed public discussion is attached to 2602.15143 yet. That is a visibility signal, not a blank module: the monitor is watching the public channels below.

Hacker News

Not indexed yet

Bluesky

Not indexed yet

PDF

Preview the source document here, or use the hero PDF action for a new tab.

References(30)

Information-Preserving Reformulation of Reasoning Traces for Antidistillation

2025Jiayu Ding, Lei Cui et al.

Virus Infection Attack on LLMs: Your Poisoning Can Spread "VIA" Synthetic Data

2025Zi Liang, Qingqing Ye et al.

Antidistillation Sampling

2025Yash Savani, Asher Trockman et al.

Do Large Language Model Benchmarks Test Reliability?

2025Joshua Vendrow, Edward Vendrow et al.

Controllable Text Generation for Large Language Models: A Survey

2024Xun Liang, Hanyu Wang et al.

MMLU-Pro: A More Robust and Challenging Multi-Task Language Understanding Benchmark

2024Yubo Wang, Xueguang Ma et al.

SemStamp: A Semantic Watermark with Paraphrastic Robustness for Text Generation

2024A. Hou, Jingyu (Jack) Zhang et al.

PromptAgent: Strategic Planning with Language Models Enables Expert-level Prompt Optimization

2023Xinyuan Wang, Chenxi Li et al.

Efficient Memory Management for Large Language Model Serving with PagedAttention

2023Woosuk Kwon, Zhuohan Li et al.

Provable Robust Watermarking for AI-Generated Text

2023Xuandong Zhao, P. Ananth et al.

Poisoning Language Models During Instruction Tuning

2023Alexander Wan, Eric Wallace et al.

Protecting Language Generation Models via Invisible Watermarking

2023Xuandong Zhao, Yu-Xiang Wang et al.

Large Language Models Are Human-Level Prompt Engineers

2022Yongchao Zhou, Andrei Ioan Muresanu et al.

Distillation-Resistant Watermarking for Model Protection in NLP

2022Xuandong Zhao, Lei Li et al.

CATER: Intellectual Property Protection on Text Generation APIs via Conditional Watermarks

2022Xuanli He, Qiongkai Xu et al.

A Comprehensive Survey on Poisoning Attacks and Countermeasures in Machine Learning

2022Zhiyi Tian, Lei Cui et al.

A comprehensive survey on robust image watermarking

2022Wenbo Wan, Jun Wang et al.

Protecting Intellectual Property of Language Generation APIs with Lexical Watermark

2021Xuanli He, Qiongkai Xu et al.

Training Verifiers to Solve Math Word Problems

2021K. Cobbe, Vineet Kosaraju et al.

FUDGE: Controlled Text Generation With Future Discriminators

2021Kevin Yang, D. Klein

Showing 20 of 30 references

CITED BY

No citing papers are indexed in the public S2S graph yet. This is an explicit zero-signal state, not a hidden lookup.

Foundation

Prior WorkDistilling LLM Feedback for Lean Theorem Proving

3.0

Prior WorkWhy Does Self-Distillation (Sometimes) Degrade the Reasoning Capability of LLMs?

3.0

Extension

Builds On ThisKnowledge Distillation Must Account for What It Loses

2.0

Commercially relevant

Higher ViabilityDistillGuard: Evaluating Defenses Against LLM Knowledge Distillation

6.0

Higher ViabilityDistillation Traps and Guards: A Calibration Knob for LLM Distillability

4.0

Higher ViabilityReinforcement-aware Knowledge Distillation for LLM Reasoning

5.0

Higher ViabilityCurriculum Learning-Guided Progressive Distillation in Large Language Models

6.0

Higher ViabilityAntidistillation Fingerprinting

6.0

Higher ViabilityHow to Steal Reasoning Without Reasoning Traces

7.0

Higher ViabilityAnti-Self-Distillation for Reasoning RL via Pointwise Mutual Information

6.0

Conflicting

none indexed

Related Resources

Why is AI security important?(question)
What is the focus of AI security research?(question)
How do AI security measures protect systems?(question)
AI Security – Use Cases(use_case)

Owned Distribution

Subscribe to the weekly brief

Get the weekly shortlist of commercializable papers, benchmark movers, and proof receipts that matter for product execution.

Agent drawer

5 surfaces preserved for agents. Humans can ignore.

Developer contracts, payload previews, evidence maps, and run controls stay here instead of the Read, Build, and Track workspace.

Run context

Paper: 2602.15143
Route: /paper/protecting-language-models-against-unauthorized-distillation-through-trace-rewriting
Active tab: read
Artifact: protecting-language-models-against-unauthorized-distillation-through-trace-rewriting

Available agents

Read extractor
Build planner
Track monitor
Competitive mapper
Related-paper scout

API/MCP endpoints

REST paper pack API/api/v1/paper/protecting-language-models-against-unauthorized-distillation-through-trace-rewriting/paper-pack
REST build passport API/api/v1/paper/protecting-language-models-against-unauthorized-distillation-through-trace-rewriting/build-passport
REST OpenAPI/api/openapi.json
MCP descriptor/api/mcp
MCP resourcesciencetostartup://surfaces/paper-workspace

Tool contracts

paper_packbuild_passportopportunity_kernelforesightsource_proofevidence_state

Payload preview

Inspect payload

{
  "contract_version": "paper-r2",
  "paper_id": "ef52e0ab-c2e5-4956-8b29-1552dd7e7d62",
  "arxiv_id": "2602.15143",
  "canonical_route": "/paper/protecting-language-models-against-unauthorized-distillation-through-trace-rewriting",
  "active_tab": "synced from current hash by the drawer client",
  "selected_artifact": "protecting-language-models-against-unauthorized-distillation-through-trace-rewriting",
  "endpoints": {
    "paper_pack": "/api/v1/paper/protecting-language-models-against-unauthorized-distillation-through-trace-rewriting/paper-pack",
    "build_passport": "/api/v1/paper/protecting-language-models-against-unauthorized-distillation-through-trace-rewriting/build-passport",
    "mcp_resource": "sciencetostartup://surfaces/paper-workspace"
  }
}

Schema validation

paper-r2 contract: present
JSON-LD twin: SSR emitted
OpenAPI path parity: /api/openapi.json
MCP resource parity: paper-workspace

Job trace

queued: drawer opened by user action
running: inspect or copy payload
succeeded: payload available in SSR
failed: route errors appear in evidence cards

Evidence map

sources used: page freshness, source proof anchors, JSON-LD
missing sources: exposed by PaperPack and EvidenceState chips
derived fallbacks: marked unverified before handoff

Page Freshness

Canonical route, proof status, last verified, refs, sources, and coverage.

Page Freshness

Paper proof surface

Canonical route: /paper/protecting-language-models-against-unauthorized-distillation-through-trace-rewriting

stale

Proof freshness: stale
Proof status: unverified
Display score: 3/10
Last proof check: 2026-04-02
Score updated: 2026-04-02
Score fresh until: 2026-05-02
References: 0
Source count: 0
Coverage: 17%

This page is showing the last landed evidence receipt and score bundle because the latest proof data is outside the freshness window.

OpenAlex: pending — this preprint is not yet indexed by OpenAlex.

Agent Handoff

Endpoint list, payload shape, route context, and copyable handoff data.

Agent Handoff

Protecting Language Models Against Unauthorized Distillation through Trace Rewriting

Canonical ID protecting-language-models-against-unauthorized-distillation-through-trace-rewriting | Route /paper/protecting-language-models-against-unauthorized-distillation-through-trace-rewriting

REST example

curl https://sciencetostartup.com/api/v1/agent-handoff/paper/protecting-language-models-against-unauthorized-distillation-through-trace-rewriting

MCP example

{
  "tool": "get_paper",
  "arguments": {
    "arxiv_id": "2602.15143"
  }
}

source_context

{
  "surface": "paper",
  "mode": "paper",
  "query": "Protecting Language Models Against Unauthorized Distillation through Trace Rewriting",
  "normalized_query": "2602.15143",
  "route": "/paper/protecting-language-models-against-unauthorized-distillation-through-trace-rewriting",
  "paper_ref": "protecting-language-models-against-unauthorized-distillation-through-trace-rewriting",
  "topic_slug": null,
  "benchmark_ref": null,
  "dataset_ref": null
}

Buildability Receipt

Verdict, compute envelope, blockers, signature state, and receipt links.

Paper proof page receipt window

Not build-ready: Protecting Language Models Against Unauthorized Distillation through Trace Rewriting

/buildability/protecting-language-models-against-unauthorized-distillation-through-trace-rewriting

Ignoreblocked

Subject: Protecting Language Models Against Unauthorized Distillation through Trace Rewriting

Verdict

Ignore

Verdict is Ignore because current viability and proof state do not clear the buildability gate.

Time to first demo

Insufficient data

No first-demo timestamp, owner estimate, or elapsed demo receipt is attached to this surface.

Compute envelope

Structured compute envelope

Insufficient data

No data, compute, hardware, memory, latency, dependency, or serving requirement receipt is attached.

Evidence ids

Receipt path

/buildability/protecting-language-models-against-unauthorized-distillation-through-trace-rewriting

Paper ref

protecting-language-models-against-unauthorized-distillation-through-trace-rewriting

arXiv id

2602.15143

Freshness

Generated at

2026-04-02T02:30:40.136Z

Evidence freshness

stale

Last verification

2026-04-02T02:30:40.136Z

Sources

References

Coverage

17%

Hash state

Lineage hash

a23a5ba065fc92d1ac0118f9755b62d7f3b6e260c3a89fc56340fc4342a2f44d

Canonical opportunity-kernel lineage hash.

Signature state

External signature

unsigned_external

No founder, registry, pilot, or production-adoption signature is attached to this receipt.

Verification

not_verified

Verification is blocked until an external signature is provided.

Blockers

Missing: repo_url
Missing: references
Missing: proof_status
Missing: distribution_readiness_scores
Missing: paper_extraction_scorecards
Unknown: distribution readiness has not been computed yet
Unknown: proof verification has not been recorded yet

Verification pending / evidence receipt incomplete

repo_url

references

Missing proof, requirement, signature, approval, adoption, or telemetry fields are blockers and must not be inferred.

Open receipt API receipt Build Loop Signal Canvas Proof divergence Divergence API Brier outcomes API

Source Proof anchors

Visual citations from the paper document graph.

JSON-LD twin

The application/ld+json payload rendered for agents.

{
  "@context": "https://schema.org",
  "@graph": [
    {
      "@type": "WebPage",
      "@id": "https://sciencetostartup.com/paper/protecting-language-models-against-unauthorized-distillation-through-trace-rewriting#webpage",
      "url": "https://sciencetostartup.com/paper/protecting-language-models-against-unauthorized-distillation-through-trace-rewriting",
      "name": "Protecting Language Models Against Unauthorized Distillation through Trace Rewriting",
      "description": "A method to modify LLM outputs to prevent unauthorized knowledge distillation and embed verifiable watermarks in student models.",
      "isPartOf": {
        "@id": "https://sciencetostartup.com/#website"
      }
    },
    {
      "@type": "ScholarlyArticle",
      "@id": "https://sciencetostartup.com/paper/protecting-language-models-against-unauthorized-distillation-through-trace-rewriting#scholarlyArticle",
      "headline": "Protecting Language Models Against Unauthorized Distillation through Trace Rewriting",
      "description": "A method to modify LLM outputs to prevent unauthorized knowledge distillation and embed verifiable watermarks in student models.",
      "url": "https://sciencetostartup.com/paper/protecting-language-models-against-unauthorized-distillation-through-trace-rewriting",
      "sameAs": "https://arxiv.org/abs/2602.15143",
      "identifier": {
        "@type": "PropertyValue",
        "propertyID": "arXiv",
        "value": "2602.15143"
      },
      "isAccessibleForFree": true,
      "isPartOf": {
        "@id": "https://sciencetostartup.com/#website"
      },
      "datePublished": "2026-02-16T19:40:07.000Z",
      "citation": [
        {
          "@type": "ScholarlyArticle",
          "identifier": {
            "@type": "PropertyValue",
            "propertyID": "SemanticScholar",
            "value": "b0de1c309d2466fd64cd237caf4d0889bfbfb47e"
          },
          "url": "https://www.semanticscholar.org/paper/b0de1c309d2466fd64cd237caf4d0889bfbfb47e"
        },
        {
          "@type": "ScholarlyArticle",
          "identifier": {
            "@type": "PropertyValue",
            "propertyID": "SemanticScholar",
            "value": "2e6d64c442138a85534912170ad9d16f08a5f6c7"
          },
          "url": "https://www.semanticscholar.org/paper/2e6d64c442138a85534912170ad9d16f08a5f6c7"
        },
        {
          "@type": "ScholarlyArticle",
          "identifier": {
            "@type": "PropertyValue",
            "propertyID": "SemanticScholar",
            "value": "73addb69df17f810bb39d4eac1ee39f031fc6f92"
          },
          "url": "https://www.semanticscholar.org/paper/73addb69df17f810bb39d4eac1ee39f031fc6f92"
        },
        {
          "@type": "ScholarlyArticle",
          "identifier": {
            "@type": "PropertyValue",
            "propertyID": "SemanticScholar",
            "value": "d1460b1a57aca30b1288ee6d6b45af25e8210bc5"
          },
          "url": "https://www.semanticscholar.org/paper/d1460b1a57aca30b1288ee6d6b45af25e8210bc5"
        },
        {
          "@type": "ScholarlyArticle",
          "identifier": {
            "@type": "PropertyValue",
            "propertyID": "SemanticScholar",
            "value": "7fbf028793a5252941e358366f57837888fcbf11"
          },
          "url": "https://www.semanticscholar.org/paper/7fbf028793a5252941e358366f57837888fcbf11"
        },
        {
          "@type": "ScholarlyArticle",
          "identifier": {
            "@type": "PropertyValue",
            "propertyID": "SemanticScholar",
            "value": "1406bb4cb6801bc4767b661308118c888a9b09da"
          },
          "url": "https://www.semanticscholar.org/paper/1406bb4cb6801bc4767b661308118c888a9b09da"
        },
        {
          "@type": "ScholarlyArticle",
          "identifier": {
            "@type": "PropertyValue",
            "propertyID": "SemanticScholar",
            "value": "1eb1a8c7f88de27af224153f43ecdd41774600f2"
          },
          "url": "https://www.semanticscholar.org/paper/1eb1a8c7f88de27af224153f43ecdd41774600f2"
        },
        {
          "@type": "ScholarlyArticle",
          "identifier": {
            "@type": "PropertyValue",
            "propertyID": "SemanticScholar",
            "value": "83b90f4a0ae4cc214eb3cc140ccfef9cd99fac05"
          },
          "url": "https://www.semanticscholar.org/paper/83b90f4a0ae4cc214eb3cc140ccfef9cd99fac05"
        },
        {
          "@type": "ScholarlyArticle",
          "identifier": {
            "@type": "PropertyValue",
            "propertyID": "SemanticScholar",
            "value": "75b68d0903af9d9f6e47ce3cf7e1a7d27ec811dc"
          },
          "url": "https://www.semanticscholar.org/paper/75b68d0903af9d9f6e47ce3cf7e1a7d27ec811dc"
        },
        {
          "@type": "ScholarlyArticle",
          "identifier": {
            "@type": "PropertyValue",
            "propertyID": "SemanticScholar",
            "value": "bf52c9d94fd61fae0d231a7e43d45d673584c282"
          },
          "url": "https://www.semanticscholar.org/paper/bf52c9d94fd61fae0d231a7e43d45d673584c282"
        },
        {
          "@type": "ScholarlyArticle",
          "identifier": {
            "@type": "PropertyValue",
            "propertyID": "SemanticScholar",
            "value": "c25d2a27f1abe169d7b68078071b6698f0980469"
          },
          "url": "https://www.semanticscholar.org/paper/c25d2a27f1abe169d7b68078071b6698f0980469"
        },
        {
          "@type": "ScholarlyArticle",
          "identifier": {
            "@type": "PropertyValue",
            "propertyID": "SemanticScholar",
            "value": "4610ffb1b016acaa82a2065ffd1a3adbae1ce722"
          },
          "url": "https://www.semanticscholar.org/paper/4610ffb1b016acaa82a2065ffd1a3adbae1ce722"
        },
        {
          "@type": "ScholarlyArticle",
          "identifier": {
            "@type": "PropertyValue",
            "propertyID": "SemanticScholar",
            "value": "c7bdd6c7b786fba60ff8a6e656cc14c3f7b57b78"
          },
          "url": "https://www.semanticscholar.org/paper/c7bdd6c7b786fba60ff8a6e656cc14c3f7b57b78"
        },
        {
          "@type": "ScholarlyArticle",
          "identifier": {
            "@type": "PropertyValue",
            "propertyID": "SemanticScholar",
            "value": "823cacd5255f3897a8d29f29a7c7cb8f978bd928"
          },
          "url": "https://www.semanticscholar.org/paper/823cacd5255f3897a8d29f29a7c7cb8f978bd928"
        },
        {
          "@type": "ScholarlyArticle",
          "identifier": {
            "@type": "PropertyValue",
            "propertyID": "SemanticScholar",
            "value": "0301917931b8e8858e7c1864e3bb73b4dc0590d2"
          },
          "url": "https://www.semanticscholar.org/paper/0301917931b8e8858e7c1864e3bb73b4dc0590d2"
        },
        {
          "@type": "ScholarlyArticle",
          "identifier": {
            "@type": "PropertyValue",
            "propertyID": "SemanticScholar",
            "value": "b3c792f282ae5d71557c2dc7f6bfe1824826cc0f"
          },
          "url": "https://www.semanticscholar.org/paper/b3c792f282ae5d71557c2dc7f6bfe1824826cc0f"
        },
        {
          "@type": "ScholarlyArticle",
          "identifier": {
            "@type": "PropertyValue",
            "propertyID": "SemanticScholar",
            "value": "2569a7309142e40815cf556b6417059df9abbda8"
          },
          "url": "https://www.semanticscholar.org/paper/2569a7309142e40815cf556b6417059df9abbda8"
        },
        {
          "@type": "ScholarlyArticle",
          "identifier": {
            "@type": "PropertyValue",
            "propertyID": "SemanticScholar",
            "value": "d6045d2ccc9c09ca1671348de86d07da6bc28eea"
          },
          "url": "https://www.semanticscholar.org/paper/d6045d2ccc9c09ca1671348de86d07da6bc28eea"
        },
        {
          "@type": "ScholarlyArticle",
          "identifier": {
            "@type": "PropertyValue",
            "propertyID": "SemanticScholar",
            "value": "b6c4a96e09b9f11e7c70e7f1fbe3f3971b92762d"
          },
          "url": "https://www.semanticscholar.org/paper/b6c4a96e09b9f11e7c70e7f1fbe3f3971b92762d"
        },
        {
          "@type": "ScholarlyArticle",
          "identifier": {
            "@type": "PropertyValue",
            "propertyID": "SemanticScholar",
            "value": "57d1e7ac339e783898f2c3b1af55737cbeee9fc5"
          },
          "url": "https://www.semanticscholar.org/paper/57d1e7ac339e783898f2c3b1af55737cbeee9fc5"
        }
      ],
      "additionalProperty": [
        {
          "@type": "PropertyValue",
          "propertyID": "viabilityScore",
          "value": 3
        },
        {
          "@type": "PropertyValue",
          "propertyID": "researchDomain",
          "value": "AI Security"
        }
      ]
    },
    {
      "@type": "BreadcrumbList",
      "itemListElement": [
        {
          "@type": "ListItem",
          "position": 1,
          "name": "Home",
          "item": "https://sciencetostartup.com"
        },
        {
          "@type": "ListItem",
          "position": 2,
          "name": "AI Security",
          "item": "https://sciencetostartup.com/topics"
        },
        {
          "@type": "ListItem",
          "position": 3,
          "name": "Protecting Language Models Against Unauthorized Distillation",
          "item": "https://sciencetostartup.com/paper/protecting-language-models-against-unauthorized-distillation-through-trace-rewriting"
        }
      ]
    }
  ]
}

Protecting Language Models Against Unauthorized Distillation through Trace Rewriting

Protecting Language Models Against Unauthorized Distillation through Trace Rewriting

Claim map

Constellation map

Competitive landscape

Buzz

PDF

References(30)

Related Papers

Related Resources

Subscribe to the weekly brief

Build artifacts

Brief

Experiment plan

Validation checklist

Scientific founder

Translational engineer

Domain operator

GTM lead

Regulatory/clinical advisor

Timeline

Claim map

Constellation map

Competitive landscape

Buzz

PDF

References(30)

Related Papers

Related Resources

Subscribe to the weekly brief

Build artifacts

Brief

Experiment plan

Validation checklist

Scientific founder

Translational engineer

Domain operator

GTM lead

Regulatory/clinical advisor

Timeline