Pith. sign in

REVIEW 12 cited by

Tamper-Resistant Safeguards for Open-Weight LLMs

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2408.00761 v4 pith:ASSWOMG3 submitted 2024-08-01 cs.LG cs.AIcs.CL

classification cs.LGcs.AIcs.CL
keywords llmsopen-weightsafeguardscapabilitiesfine-tuningmethodstepstamper-resistance
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Rapid advances in the capabilities of large language models (LLMs) have raised widespread concerns regarding their potential for malicious use. Open-weight LLMs present unique challenges, as existing safeguards lack robustness to tampering attacks that modify model weights. For example, recent works have demonstrated that refusal and unlearning safeguards can be trivially removed with a few steps of fine-tuning. These vulnerabilities necessitate new approaches for enabling the safe release of open-weight LLMs. We develop a method, called TAR, for building tamper-resistant safeguards into open-weight LLMs such that adversaries cannot remove the safeguards even after hundreds of steps of fine-tuning. In extensive evaluations and red teaming analyses, we find that our method greatly improves tamper-resistance while preserving benign capabilities. Our results demonstrate that progress on tamper-resistance is possible, opening up a promising new avenue to improve the safety and security of open-weight LLMs.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 12 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Emergent Misalignment Recruits a Pre-existing Persona Subspace

    cs.LG 2026-07 conditional novelty 7.0 of 10

    Fine-tuning on narrow bad data recruits a low-rank persona subspace already present in a frozen instruction-tuned model; holding that subspace out of activations prevents broad misalignment, and injecting it into the ...

  2. Breaking the Rounding Trap: Securing LLMs against Quantization-Conditioned Backdoors

    cs.CR 2026-06 unverdicted novelty 6.0 of 10

    QuantGuard uses differentiable rounding control to break quantization-boundary backdoors in LLMs, reducing post-quantization attack success to clean-model levels across six models and INT8/FP4/NF4.

  3. RippleBench: Capturing Ripple Effects Using Existing Knowledge Repositories

    cs.AI 2025-12 conditional novelty 6.0 of 10

    RippleBench automatically generates questions at increasing semantic distance from unlearned topics and shows all eight tested unlearning methods degrade accuracy that recovers only slowly with distance.

  4. Towards Integrated Alignment

    cs.CY 2025-08 conditional novelty 6.0 of 10

    LLM-generated textbook-style forget sets, produced from a domain name alone, achieve unlearning performance comparable to expert-curated datasets in biosecurity, cybersecurity, and Harry Potter benchmarks.

  5. The Safety Gap Toolkit: Evaluating Hidden Dangers of Open-Source Models

    cs.CY 2025-07 conditional novelty 6.0 of 10

    On Llama-3 and Qwen-2.5, removing safety guardrails sharply raises compliance with dangerous bio, chem, and cyber requests, and the resulting safety gap grows with model scale.

  6. FORTRESS: Frontier Risk Evaluation for National Security and Public Safety

    cs.CY 2025-06 conditional novelty 6.0 of 10

    A new benchmark with instance-specific rubrics measures frontier LLMs' willingness to assist with national security and public safety threats, alongside a paired over-refusal test.

  7. Existing Large Language Model Unlearning Evaluations Are Inconclusive

    cs.LG 2025-05 conditional novelty 6.0 of 10

    Existing LLM unlearning evaluations are inconclusive: they can inject new information, depend heavily on task format, and rely on spurious correlations.

  8. Engineering Trustworthy Agentic AI for Critical Systems

    cs.AI 2026-07 conditional novelty 5.0 of 10

    A survey claiming that agentic AI trustworthiness is a single cross-domain problem and outlining a framework for graded, certifiable assurance.

  9. Technical Requirements for Halting Dangerous AI Activities

    cs.AI 2025-07 conditional novelty 5.0 of 10

    A taxonomy of compute-centric technical interventions, graded by readiness and mapped to five AI governance plans, argues that halting dangerous AI requires substantial control over AI compute.

  10. Prompt Attacks Reveal Superficial Knowledge Removal in Unlearning Methods

    cs.CR 2025-06 conditional novelty 5.0 of 10

    Prepending a Hindi filler paragraph to WMDP-bio questions restores 57.3% accuracy in ELM-unlearned models, showing the unlearning is superficial output suppression rather than true knowledge removal.

  11. LLM Unlearning for Cyber Defense: A Survey on Methods, Challenges, and Emerging Threats

    cs.LG 2026-06 conditional novelty 4.0 of 10

    Most gradient-based LLM unlearning methods achieve behavioral suppression, not true forgetting, and current benchmarks cannot certify that knowledge has been removed.

  12. Step-by-Step Reasoning Attack: Revealing 'Erased' Knowledge in Large Language Models

    cs.CR 2025-06 reject novelty 4.0 of 10

    Step-by-step reasoning prompts can recover purportedly erased facts from unlearned LLMs, but the paper's quantitative evidence is internally inconsistent.

Pith tools