Chatter Attack drives LLMs to responses up to 31.5 times longer
Prompts target short doubt-inducing phrases, and Bayesian Prompt Optimization extends the attack to black-box models.
Stories per day · 30 days
2 RSS
Relateddenial-of-service2adversarial-prompts1document-processing1json1llm-inference1progress1proof-of-concept1radwordsprocessing1struts21telerik1
6 stories, newest first · one card per story
Stories tagged #resource-consumption
JSON array proof of concept shows parser resource exhaustion
The authors say Apache Struts2 JSON Plugin changes through S2-072 still do not stop maximised comma-array consumption.
Telerik Document Processing before 2026.3.1027 can exhaust resources (CVE-2026-118035)
Canada's Cyber Centre points administrators to Progress's advisory for RadWordsProcessing and says to apply updates when available.
CVE-2026-104712: Apache Struts: Disproportionate response size when rendering BigDecimal request parameters
Apache Struts can amplify responses when rendering BigDecimal parameters, tracked as CVE-2026-104712.
FragToken: Amplifying LLM Inference Costs through Noncanonical Token Generation
FragToken trains LLMs to emit longer non-canonical tokens, roughly doubling inference cost while keeping answers usable.
CVE-2026-77791: Apache Tomcat: DoS via busy wait during WebSocket close
Apache Tomcat WebSocket close handling can busy-wait, enabling denial of service across multiple supported versions.