DeGenTWeb: A First Look at LLM-dominant Websites
2026-05-04T07:23:59Z•656627a3984c1073196b33a2e55cd42c98cbaeb2a98088d9aeb35ef53b23719a
BGPDeGenTWebIranLLM routingLLM-generated contentMixture-of-ExpertsMoE servingQ-GUARDcost-effectivenessdata removaldetectorsdigital twinsentanglement routinginternet censorshipmodel profilesnetwork managementnetwork topologiesnull-routingoutage detectionprivacy-compliancequantum networkingrouting systemsswitchless networksuntwinningweb measurement
What happened
This batch of arXiv items spans networked-systems, ML/LLM infrastructure, and privacy/control mechanisms. Key contributions: DeGenTWeb — systematic method to identify LLM-dominant websites, showing high and growing prevalence in Common Crawl and Bing and noting that off-the-shelf LLM detectors underperform in realistic web settings; Network Digital Untwinning — mechanisms (single and parallel) for targeted removal of contributions from network digital twins with rollback+noise and remapping and provable indistinguishability; RouteProfile — a structured design space for LLM profiles that shows
Why it matters
A reviewed impact interpretation has not been published for this record.
Evidence and limitations
- Source ID
- arxiv_cs_ni
- Record identifier
- 656627a3984c1073196b33a2e55cd42c98cbaeb2a98088d9aeb35ef53b23719a
- Enrichment time
- 2026-05-04T07:23:59Z
- 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.