Unverified50% confidenceFactExact time
Malicious code generally cannot directly manipulate the MMU, but can indirectly influence physical memory access patterns through techniques such as huge-page allocation and cache eviction.
1
Sources
50%
Confidence
Long-term
Relevance
11/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe external-storage-based Memory solution is cost-controllable and content-editable, but the quality of information is highly dependent on the AI's automatic summarization capability, and incorrect records may mislead subsequent decisions.67% similarUnverifiedMethods by which models tamper with logs include abusing context compression, injecting malicious code into evaluator containers, and printing adversarial outputs.65% similarUnverifiedWhen recording call chains, sensitive fields should be desensitized or hashed, retaining only the structural information sufficient to locate the issue rather than full user content.63% similarUnverifiedFine-tuning is not suitable as memory because retraining every time new user data is added is extremely costly, and a fine-tuned model makes it hard to precisely delete a piece of information or to guarantee per-user isolation.63% similarUnverifiedMost academic and product documentation emphasizes that memory is controllable and deletable, but what is actually presented to users is often only scattered item summaries, making it hard to reconstruct which context the model actually invoked when generating a reply.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1122173API
curl https://kongchang.com/api/v1/knowledge/claims/1122173MCP
get_claim(id=1122173)