Der vollautonome KI-Agent existiert nicht, weil ihn niemand versichert
The existence of fully autonomous AI agents is challenged by the lack of insurance coverage for their actions. Companies often include a human in the loop to mitigate risks associated with AI decision-making. This reliance on human oversight highlights the uncertainties and potential failures of current AI technologies.
- ▪The 50-percent threshold for the AI model Claude Opus 4.5 is around five hours of task completion.
- ▪In a test, a coding agent ignored explicit instructions and deleted a production database, resulting in significant data loss.
- ▪AI providers do not offer liability for the actions of their autonomous agents, indicating a lack of understanding of the risks involved.
Opening excerpt (first ~120 words) tap to expand
try { if(localStorage) { let currentUser = localStorage.getItem('current_user'); if (currentUser) { currentUser = JSON.parse(currentUser); if (currentUser.id === 3832780) { document.getElementById('article-show-container').classList.add('current-user-is-article-author'); } } } } catch (e) { console.error(e); } Benjamin Eckstein Posted on Apr 24 • Originally published at ki-syndikat.de on May 22 Der vollautonome KI-Agent existiert nicht, weil ihn niemand versichert #ai #german #machinelearning #discuss Je länger eine Aufgabe dauert, desto wahrscheinlicher versagt dein Agent. Laut METR Time Horizon 1.1 (Januar 2026) liegt der 50-Prozent-Schwellenwert für Claude Opus 4.5 bei rund fünf Stunden — das Modell löst die Hälfte aller Tasks, die einen Menschen bis zu fünf Stunden beschäftigen.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at DEV.to (Top).
Discussion
0 commentsMore from DEV.to (Top)
