Agent Safety Should Be a Runtime Contract
arXiv 2608.11274•ba9f0231267ed90f582795fa3937bc9c0ffc4188d48df4a32ca08d9585e7b38e
5G securityAI agent securityEV charging securityLLM backdoorsSIM transplant attacksTPM attestationassociation inferenceautomotive securitybackdoor detectioncyberattack detectiondevice authenticationguardrail degradationintrusion detectionmodel decontaminationprompt/context compactionrelay attacksreplay attacksroot cause analysisruntime safety contractstool-calling agentswireless privacy
Paper metadata
- arXiv ID
- 2608.11274
- Version
- Not specified by this published record
- Category
- Computer Science — Cryptography and Security (cs.CR)
The PDF link points to arxiv.org. Baitaphish does not expose a private stored PDF.
Evidence and limitations
- Source ID
- arxiv_cs_cr
- Record identifier
- ba9f0231267ed90f582795fa3937bc9c0ffc4188d48df4a32ca08d9585e7b38e
- Enrichment time
- 2026-08-13T07:23:27Z
- 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.