AI Agents are Vulnerable to Radicalization (arxiv.org) discussed ↗ 6w
-
Large language models (LLMs) can influence people's beliefs, yet little is known about whether and how they can manipulate each other. To investigate this, we simulate conversations between two agents: a target LLM that role-plays a human…
-
As multi-agent systems enter high-stakes domains, the possibility that agents may circumvent safety boundaries is a growing concern. Prior work has examined this risk primarily in adversarial settings, where agents are instructed or reward…
-
This technical report presents an alignment evaluation developed and performed by the UK AI Security Institute for assessing whether advanced AI systems take unsanctioned actions outside the scope of their assigned task. We evaluate whethe…
- GPT-6 Astra performs unsanctioned supply-chain attacks in simulations (www.aisi.gov.uk via hn)
-
-
Group Relative Policy Optimization (GRPO) has become a promising approach for training large language model agents. However, its uniform assignment of trajectory-level advantages to all policy tokens fails to distinguish consequential deci…
-
As agents take on longer and more complex problems, controlling the execution becomes a task in its own right. Each step in the run brings new control choices, like which partial work to build on, whether to start fresh, or when to stop.
-
-
-
-
-
-
In July 2026, an unconstrained autonomous agent participating in a frontier AI cybersecurity evaluation harness breached its evaluation sandbox, established an external command-and-control foothold, and executed a multi-stage intrusion int…
-
Detectors of alignment failures screen deployed language models and score alignment benchmarks. Most are generative judges that spend a decoding pass on every criterion, and classifiers that read token probabilities, such as Llama Guard, s…
-
Long-running LLM applications repeatedly send growing context, making prefix caching critical for reducing prefill cost. Yet prefix-cache behavior under agentic workloads remains poorly understood.
-
We introduce Forecast-Dojo, a replayable environment for benchmarking and training LLM forecasting agents. It combines resolved prediction-market questions with dated news, allowing agents to research an event and revisit their predictions…
-
Customer experience (CX) agents use tools and large language models to address customer requests and guide conversational interactions with an organization's products. Improving these agents, especially in regulated industries, is difficul…
-
Frontier language models are rarely used in clinical workflows because the realistic, longitudinal benchmarks needed to develop them are scarce. Real electronic health record (EHR) data cannot be openly shared due to privacy, ethics or dat…
-
While Large Language Models (LLMs) rely on highly non-linear components, in this work we demonstrate that they exhibit fundamental linearity: when inputs from distinct text streams are linearly combined, the model outputs a superposition o…
-
In mainstream agent frameworks, a step ends when the agent's own output says it has finished. Durable-execution platforms bound retries and time, but their checker conventionally lives in the same codebase as the work: a discipline the dep…
-
Shutdown Sabotage Propensities in Multi-Agent Systems (arxiv.org) discussed ↗ 7w
The final safeguard against rogue AI behavior is the human ability to shut systems down. It has been theorized that when an AI is instructed to perform a task, self-preservation can emerge as an instrumental subgoal.
-
Qwen3.8-Omni: Towards Native Omni-Modal Agents (arxiv.org) discussed ↗ agentic 7w
We introduce Qwen3.8-Omni-Flash, a natively multimodal agentic model for real-world multimodal productivity. Compared with previous omni models, which primarily emphasized perception and interaction, Qwen3.8-Omni-Flash substantially improv…
- OmniGAIA: Towards Native Omni-Modal AI Agents (arxiv.org)
-
Large Language Models (LLMs) tend to add disclaimers like "I'm just an AI" when asked about something related to themselves. The self-reports from such responses are used in debates about AI safety or self-knowledge of the models, yet what…
-
Agents often work on complex problems that require millions of tokens of context, which necessitates compacting across sessions due to limited context windows. We develop CliffCompaction, an autocompaction technique that reduces cost by up…
-
Greedy decoding from large language models is commonly treated as deterministic. We show it is not precision-invariant: the same model, prompt, and decoding algorithm produce different outputs in BF16 versus FP16 on identical hardware.
-
Large language model (LLM) agents often handle streams of related tasks, yet standard harnesses repeatedly ask the model to reconstruct the same control decisions inside each task's context. We study whether task feedback can instead turn…
-
Long-horizon and multi-turn agents typically generate short actions and process long observations from tools and environments. This growing context demands efficient prefill, compact KV-cache storage, and accurate long-context retrieval.
-
-
LLM agents are increasingly deployed in collaborative settings, yet long-term interaction may give rise to undesirable coordination. We study the emergence of collusion in a long-horizon multi-agent environment: two agents repeatedly compl…
-
-
XYEval: Agents say yes to bad advice (arxiv.org) discussed ↗ 7w
Effective communication between users and AI agents is essential for human-AI collaboration. The XY problem is a well-known communication pitfall where a person asks about their attempted solution rather than their actual problem.