attention head AI jobs Alerts Saved

Jobs by paper

Work with the people writing the research. 819 papers from 61 employers hiring for AI roles now, each beside its employer's open roles, straight from their own job board.

35 papers by Anthropic. Anthropic's page has its open roles.

PaperEmployerPublishedCitedOpen roles
The Memory Trap: Language Agents Overgeneralize from Benign InteractionsarXiv · authors at Anthropic Anthropic Nov 2026 1 633
Agentic Backdoors via Pretraining PoisoningarXiv · authors at Anthropic Anthropic Oct 2026 1 633
Model Organisms of Collusion: Corrective Transfer in Untrusted MonitorsarXiv · authors at Anthropic Anthropic Oct 2026 0 633
Probabilistic Sensing, Deterministic Authority: Admitting Model-Produced Observations into Sufficiency-Checked Governance ContractsarXiv · authors at Anthropic Anthropic Oct 2026 0 633
A Bird's-Eye View of Iterative Reward DesignarXiv · authors at Anthropic Anthropic Oct 2026 0 633
When Do Models Admit They Are Wrong? Failure Disclosure Is Unstable Under Reinforcement LearningarXiv · authors at Anthropic Anthropic Sep 2026 0 633
Improving Evaluation Realism with Inference-Time Compute and Deployment ScaffoldsarXiv · authors at Anthropic Anthropic Sep 2026 0 633
Mind Viruses: Self-Propagating Ideas in Multi-Agent LLM SystemsarXiv · authors at Anthropic Anthropic Aug 2026 10 633
A Shared Subcircuit Lets LLMs Count Down Across TasksarXiv · authors at Anthropic Anthropic Jul 2026 1 633
The Value Axis: Language Models Encode Whether They're on the Right TrackarXiv · authors at Anthropic Anthropic Jun 2026 9 633
Diffuse AI Control on Fuzzy TasksarXiv · authors at Anthropic Anthropic Jun 2026 2 633
(Mis)generalization of Helpful-Only Fine-tuningarXiv · authors at Anthropic Anthropic Jun 2026 2 633
From Simulation to Enaction: Post-trained language models recognize and react to their own generationsarXiv · authors at Anthropic Anthropic May 2026 5 633
Faithfulness as Information Flow: Evaluating and Training Faithful Chain-of-Thought ReasoningarXiv · authors at Anthropic Anthropic May 2026 7 633
SLEIGHT-Bench: A Benchmark of Evasion Attacks Against Agent MonitorsarXiv · authors at Anthropic Anthropic May 2026 11 633
Removing Sandbagging in LLMs by Training with Weak SupervisionarXiv · authors at Anthropic Anthropic Apr 2026 6 633
Language models transmit behavioural traits through hidden signals in dataNature · authors at Anthropic Anthropic Apr 2026 191 633
Mechanisms of Introspective AwarenessarXiv · authors at Anthropic Anthropic Mar 2026 19 633
Gemma Needs Help: Investigating and Mitigating Emotional Instability in LLMsarXiv · authors at Anthropic Anthropic Feb 2026 14 633
The Assistant Axis: Situating and Stabilizing the Default Persona of Language ModelsarXiv · authors at Anthropic Anthropic Jan 2026 148 633
Excess Description Length of Learning Generalizable PredictorsarXiv · authors at Anthropic Anthropic Jan 2026 2 633
Activation Oracles: Training and Evaluating LLMs as General-Purpose Activation ExplainersarXiv · authors at Anthropic Anthropic Dec 2025 70 633
Beyond Data Filtering: Knowledge Localization for Capability Removal in LLMsarXiv · authors at Anthropic Anthropic Dec 2025 17 633
Verifying LLM Inference to Detect Model Weight ExfiltrationarXiv · authors at Anthropic Anthropic Nov 2025 10 633
Optimizing AI Agent Attacks With Synthetic DataarXiv · authors at Anthropic Anthropic Nov 2025 8 633
Believe It or Not: How Deeply do LLMs Believe Implanted Facts?arXiv · authors at Anthropic Anthropic Oct 2025 40 633
Poisoning Attacks on LLMs Require a Near-constant Number of Poison SamplesarXiv · authors at Anthropic Anthropic Oct 2025 166 633
Inoculation Prompting: Instructing LLMs to misbehave at train-time improves test-time alignmentarXiv · authors at Anthropic Anthropic Oct 2025 52 633
Agentic Misalignment: How LLMs Could Be Insider ThreatsarXiv · authors at Anthropic Anthropic Oct 2025 324 633
Eliciting Secret Knowledge from Language ModelsarXiv · authors at Anthropic Anthropic Oct 2025 41 633
Inverse Scaling in Test-Time ComputeEdinburgh Research Explorer (University of Edinburgh) · authors at Anthropic Anthropic Jul 2025 78 633
Teaching Models to Verbalize Reward Hacking in Chain-of-Thought ReasoningarXiv · authors at Anthropic Anthropic Jun 2025 28 633
Reasoning Models Don't Always Say What They ThinkarXiv · authors at Anthropic Anthropic May 2025 719 633
Constitutional Classifiers: Defending against Universal Jailbreaks across Thousands of Hours of Red TeamingarXiv · authors at Anthropic Anthropic Jan 2025 286 633
Subversion Strategy Eval: Can language models statelessly strategize to subvert control protocols?arXiv · authors at Anthropic Anthropic Dec 2024 30 633

Papers of the last two years, tied to employers on evidence only: papers an employer's Hugging Face organisation lists as its own, and papers whose authors give the employer as their affiliation in OpenAlex. A paper's roles are its employer's: postings rarely say which team wrote what. How papers are matched.