Blockchain Attacks and Defenses: A Layered and Cross-Domain Survey
2026-07-09T07:23:31Z•7d785515710ed89934773cb1ad05431e584f8dd0993b5cc8d44f413c0cf0f160
accountability-protocolsagentic-aiam-sentrybackdoor-absorptionbackdoor-attackscalibration-family-overfitcontradiction-proofsdecentralized-trainingdrI-xappseco-cpo-dagfederated-learningfully-homomorphic-encryptionghostwritermemory-poisoningml-securitymonitor-transferabilitymulti-agent-systemsopen-ranoran-defendpopsprivacyproveflsupply-chain-securityunlearningverifiable-aggregation
What happened
Collection of new security/privacy research covering high-impact vulnerabilities and defenses across blockchain, agentic AI, federated learning, and ML supply chains. Key findings: GhostWriter — a memory-poisoning attack against long-term-memory, tool-using personal agents with ~98% injection and ~60% activation rates (proposed mitigations: AM-Sentry). POPS — a prompt-optimized parameter-shaking adversary that can recover supposedly unlearned multimodal private data from MLLMs. Several works describe backdoor threats (decentralized/community training, DRL xApps in O-RAN) and practical defenses
Why it matters
A reviewed impact interpretation has not been published for this record.
Evidence and limitations
- Source ID
- arxiv_cs_cr
- Record identifier
- 7d785515710ed89934773cb1ad05431e584f8dd0993b5cc8d44f413c0cf0f160
- Enrichment time
- 2026-07-09T07:23:31Z
- 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.