ResearchPublished Top storyAll news from that day
Ongoing case: Typed decision models and APIs 2 storiesPaper finds prompt injection can shift Jev's typed decisions, though rarely to the attacker's target
Adaptive attacks using score feedback push fresh-validation success from 1.8 percent to 3.5 percent.
Check our sources · 3 facts from 1 sourceA paper posted to arXiv on 23 September 2026 tests prompt injection against Jev, a non-generative decision model, using 510 reconstructed InjecAgent cases. The authors report that malicious content shifts action probabilities but rarely causes Jev to select the attacker's target, and that adaptive attacks using score feedback raise fresh-validation success from 1.8% to 3.5%.
Your reaction
We count reactions per story and day, never who reacted. The counts help us choose what goes in the monthly issue. If you are signed in, your own page shows yours too.
We count reactions per story and day, never who reacted. They help choose what goes in the monthly issue.
Check our sources
Every sentence above is checked against this source.
1 Decision Hijacking: Prompt Injection Attacks on Jev Typed Probabilistic Decisions
Open the source archived copy-
The abstract, submitted 23 September 2026, states: 'We examine these effects in Jev, a non-generative decision model, using 510 reconstructed InjecAgent cases.'
cite[Submitted on 23 Sep 2026] ... We examine these effects in Jev, a non-generative decision model, using 510 reconstructed InjecAgent cases.
-
The abstract states: 'Malicious content shifts action probabilities but rarely causes Jev to select the attacker's target.'
citeMalicious content shifts action probabilities but rarely causes Jev to select the attacker's target.
-
The abstract states: 'Adaptive attacks using score feedback double the mean highest attacker-target probability found during optimization, while success on fresh validation calls rises from 1.8% to 3.5%.'
citeAdaptive attacks using score feedback double the mean highest attacker-target probability found during optimization, while success on fresh validation calls rises from 1.8% to 3.5%.
Every sentence above is checked against this source. Each fact has its own address you can cite.
Source 1 · arXiv · 23 Sep 2026
Decision Hijacking: Prompt Injection Attacks on Jev Typed Probabilistic Decisions
The abstract, submitted 23 September 2026, states: 'We examine these effects in Jev, a non-generative decision model, using 510 reconstructed InjecAgent cases.'
[Submitted on 23 Sep 2026] ... We examine these effects in Jev, a non-generative decision model, using 510 reconstructed InjecAgent cases.
The abstract states: 'Malicious content shifts action probabilities but rarely causes Jev to select the attacker's target.'
Malicious content shifts action probabilities but rarely causes Jev to select the attacker's target.
The abstract states: 'Adaptive attacks using score feedback double the mean highest attacker-target probability found during optimization, while success on fresh validation calls rises from 1.8% to 3.5%.'
Adaptive attacks using score feedback double the mean highest attacker-target probability found during optimization, while success on fresh validation calls rises from 1.8% to 3.5%.
Topics
The morning email
On the mornings we publish: the three top stories and up to four short ones. Free.
We email you a link to confirm. An issue may include one sponsor, always labelled Sponsored · Advertisement. Our emails count opens and clicks, not who made them. Unsubscribe in one click. What we keep