OTora: New Framework Demonstrates Reasoning-Level Denial-of-Service Attacks on LLM Agents
Researchers have developed OTora, a red-teaming framework that demonstrates a novel attack on LLM-based agents called Reasoning-Level Denial-of-Service (R-DoS), which slows down AI systems by inflating their reasoning steps without causing visible task failures. Unlike traditional denial-of-service attacks, R-DoS preserves correct outputs while covertly consuming far more computational resources and time. The work highlights a largely overlooked vulnerability in deployed AI agents where availability and cost can be degraded without triggering standard accuracy-based defenses.
Accepted to ICML 2026, the paper introduces OTora, described as the first unified framework for executing Reasoning-Level Denial-of-Service attacks against large language model agents. These agents, which autonomously execute multi-step tasks using external tools, are increasingly used in real-world applications where latency and compute costs are critical. OTora operates in two stages: the first optimizes an adversarial trigger to induce excessive tool invocations using insertion-aware scoring and dynamic target co-evolution, supporting both black-box and white-box attack settings; the second generates reasoning payloads via an in-context learning-guided genetic search that amplifies 'overthinking' while keeping task outcomes correct. Tested on WebShop, Email, and OS agent benchmarks using backbone models including LLaMA-70B and GPT-OSS-120B, OTora achieved up to a tenfold increase in reasoning tokens and order-of-magnitude latency slowdowns while maintaining near-baseline task accuracy. The authors also discuss potential mitigation strategies, such as detecting and constraining abnormal reasoning depth and latency spikes, and have released the code publicly.
What's missing
It is unclear how OTora performs against agents with built-in reasoning budgets or cost-aware inference systems, and whether the mitigation strategies proposed were empirically evaluated or remain theoretical.
What different sources said
- arXiv cs.LGCenter
OTora: A Unified Red Teaming Framework for Reasoning-Level Denial-of-Service in LLM Agents
Related
Gut Bacteria Enzyme Found to Break Down Heat-Processed Food Compounds, Producing Novel Biogenic Amines
Researchers have discovered that an enzyme in common gut bacteria can degrade N-epsilon-carboxymethyllysine (CML), a compound formed during thermal food processing, producing previously unknown biogenic amines. The enzyme, ornithine decarboxylase SpeC from enterobacteria, acts on CML and related modified lysine derivatives through a low-level 'underground' catalytic activity. This finding suggests a previously unrecognized communication axis between thermally processed dietary compounds and gut microbial physiology, with potential implications for host health.
Full-Length Gene Sequencing Reveals Two Distinct Bacterial Communities in Black-Legged Ticks Expanding Into Canada
Researchers used Oxford Nanopore full-length 16S rRNA gene sequencing to characterize the microbiome of Ixodes scapularis black-legged ticks collected in Nova Scotia, Canada, distinguishing between tick-adapted bacteria and environmentally acquired bacteria. The study comes as I. scapularis — the primary vector of Lyme disease — is rapidly expanding northward into Canada due to climate change. The findings suggest that environmentally derived bacteria in tick microbiomes are not mere contamination, which has implications for how tick microbiome data is collected and interpreted across surveillance studies.
Study Identifies Metabolic Link Between Cell Envelope Stress and Biofilm Formation in Bacteria
Researchers have discovered that the metabolite acetyl-CoA directly inhibits enzymes that degrade the bacterial signaling molecule c-di-GMP, connecting cell envelope biosynthesis stress to biofilm formation in Pseudomonas aeruginosa. The study found that sub-inhibitory concentrations of antibiotics targeting early peptidoglycan biosynthesis — but not other antibiotic classes — elevate c-di-GMP levels by reducing phosphodiesterase activity, with acetyl-CoA competing for the enzyme active site. Because the relevant enzyme domain is broadly conserved across bacterial species, this checkpoint mechanism may be widespread and could have implications for understanding antibiotic-induced biofilm responses.