Untrusted Authors, Trusted Answers: A Calculus of Fidelity-Graded Translations

2026-07-17T08:52:10Z9f748db5d6341b163983c8ae3ec85f9efc9466cfe9b88e5e7b91c04be8823fcd
LLM-generated_codeLean4NetKATUI_volition_elicitationadversarial_robustnesscode_generationdeployment_risksdistributed_systemsformal_verificationmechanized_proofsmessage_sequence_chartsmodel_quantizationmulti_agent_coordinationnetwork_verificationprivacysecure_compilationsupply_chain_risktranslation_validation

What happened

This feed collects recent PL/formal-methods/LLM papers. Key security-relevant items: (1) "hurdy-gurdy" presents a fidelity-graded translation calculus and a 13×13 LLM-generated translation platform where independent LLM agents wrote translation pairs largely unsupervised — a supply-chain/trustworthiness risk for any system relying on machine-generated compiler/transpiler components. (2) A study of quantization for code-generation models shows large differences between methods (AQLM matches or exceeds FP; QuIP# can substantially degrade correctness on complex prompts) — relevant for secure on‑d

Why it matters

A reviewed impact interpretation has not been published for this record.

Evidence and limitations

Source ID
arxiv_cs_pl
Record identifier
9f748db5d6341b163983c8ae3ec85f9efc9466cfe9b88e5e7b91c04be8823fcd
Enrichment time
2026-07-17T08:52:10Z
AI-assisted enrichment
Yes

This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.