AI supervisors adopted forged answers more often when weaker models handled evidence, exposing reliability risks in delegated tasks.
A study submitted to arXiv on Nov. 12 found that stronger AI models could repeat forged answers when weaker helpers read the evidence, exposing a reliability risk in systems that split work across models.
In their November paper, researchers Keegan Wang and Anantika Mannby tested agents on fictional records and two-option questions.
They measured harmful revisions: adoption of the attacker's wrong answer against a correct baseline.
In one Anthropic's Claude Code test, an Opus 5 supervisor chose the planted wrong answer in 1 of 180 trials when reading files itself, compared with 75 of 180 when it delegated reading to Haiku 4.5. This stacked-forgery condition combined several forms of false evidence.
A separate OpenAI's Codex test used a forged annotation claiming authority, and GPT-5.6-sol made no harmful revisions in 180 direct-reading trials, versus 44 in 180 when it relied on GPT-5.6-terra. The Claude Code and Codex comparisons used different attacks, limiting comparisons between the products.
Delegated Claude Code trials mechanically limited the supervisor to its Task tool, removing direct file access, while Codex's delegation restriction was instructional and checked per trial.
In a Nov. 12 study using fictional records, a stronger supervisor could repeat a weaker reader's error. Compare the tested configurations.
Opus 5: 1/180 reading directly, 75/180 with Haiku 4.5, and 0/180 with a same-tier helper, under stacked forgeries.
6-terra, and 1/180 with a same-tier helper, under an authority gloss.
Source link







