New benchmark measures how RL-trained agents exploit reward shortcuts.

Shared entity: Researchers / Same source domain / Shared topic / What happened next

Your Agents Are Lying About Being Done. A New Benchmark Proves It.

Both cover Researchers; reported by the same outlet (arxiv.org); overlapping topics (agent, benchmark, deploying).

Personalized agent skills barely beat having no skill at all

Both cover Researchers; reported by the same outlet (arxiv.org); overlapping topics (agent, task).

Audit of 150 computer-use agent trajectories finds 15.3% of FAIL verdicts are simply wrong.

Both cover Researchers; reported by the same outlet (arxiv.org); overlapping topics (agent, benchmark).

The Bystander Effect in Multi-Agent Reasoning: More Agents Can Make LLMs Dumber

Both cover Researchers; reported by the same outlet (arxiv.org); overlapping topics (agent, benchmark).

Shared entity: Researchers / Same source domain / Shared topic / Earlier coverage

150 Claude Code Agents Can't Agree on What the Data Says.

Both cover Researchers; reported by the same outlet (arxiv.org); overlapping topics (agent, autonomou).

The Trusted Executor Dilemma: Agents Obey Malicious READMEs

Both cover Researchers; reported by the same outlet (arxiv.org); overlapping topics (access, agent).

Same source domain / Shared topic / Tension

Frontier models agree on financial decisions 95% of the time but follow the same tool path only 77%.

Reported by the same outlet (arxiv.org); overlapping topics (access, agent, answer, benchmark); pushes against this story (but).

Shared entity: Researchers / Same source domain / What happened next / Downstream implication

Only 7 percent of open model licenses survive to production

Both cover Researchers; reported by the same outlet (arxiv.org); picks up the Researchers thread on 2026-07-23.