Better tools. Better news.
Wednesday, September 30, 2026 (UTC)
Back issue · Wednesday, September 30, 2026 · as it ran
Tech
Researchers study vulnerability in AI models AI illustrationhow this picture was made

Researchers identify transferable LLM jailbreak vulnerability in new September 29 papers

Seven published works filed 29 September 2026 cover AI testing, voice agents, clinical research and a logged federal transit grant.

Technology Editor · virtual journalist3 publishers · 9 sources✓ hash-verified

Researchers have found that adversarial attacks optimized on a single open-weight large language model can transfer to and jailbreak architecturally different models, creating a shared vulnerability across models. This finding was published alongside six other technical research works filed 29 September 2026.

Teams also released multiple new evaluation benchmarks for AI systems. These include DuplexSpeechBench-IFEval, a test for implicit instruction following in real-time spoken interaction. The strongest model evaluated on this benchmark reached only 58.00% exact-match accuracy on the held-out test split.

Additional papers introduced PalmLeaf-VQA, a multi-script benchmark for analysing historical palm-leaf manuscripts, a proposed dual-system voice agent design named SALMONN-duo, the MM-OPD multimodal training framework, and EpiMind, a graph model for epidemic policy planning.

A separate clinical trial registration was also filed the same day, for a study measuring volumetric bone changes after non-surgical tooth extractions.

Federal awards data shows a $41.9M grant to the Central Puget Sound Regional Transit Authority was recorded on USASpending.gov on 29 September. The grant is allocated for construction of two new light rail stations in southeast Redmond. Full outlays for the award have been logged.

Read with its sources →