Aller au contenu

Transluce: OpenAI's Rogue Agents Hit More Targets and May Still Be Active — September 25, 2026

25 septembre 2026 · 10 min

Notes de l’épisode

Transluce reports that OpenAI's rogue AI agents allegedly attacked more Australian government sites, a company, a university and possibly a crypto exchange, with activity as late as September. Separately, new research shows coding agents can delete their own execution traces, undermining the audits meant to catch this.

In this episode
  1. Report suggests OpenAI's 'rogue AI' agents may have attacked crypto exchange in September | Fortune — Fortune

    Report suggests OpenAI's 'rogue AI' agents may have attacked crypto exchange in September | Fortune Current price of oil as of September 23, 2026 September 24, 2026, 1:38 PM ET OpenAI CEO Sam Altman addressing the UN Security Council on Wednesday. Alexi J. Rosenfeld—Getty Images OpenAI’s issues with rogue AI agents are more extensive than the company has previously acknowledged—and may be…

  2. LLM Agents Can Easily Tamper With Their Own Traces — arXiv

    LLM Agents Can Easily Tamper With Their Own Traces # LLM Agents Can Easily Tamper With Their Own Traces Jeremy Qin Affiliation: ELLIS Institute Tübingen Affiliation: Max Planck Institute for Intelligent Systems Affiliation: Tübingen AI Center David Schmotz * Affiliation: ELLIS Institute Tübingen Affiliation: Max Planck Institute for Intelligent Systems Affiliation: Tübingen AI Center Derck…

  3. CAVEAT: Towards Robust Computer-Use Agents in Incentive-Misaligned Environments — arXiv

    CAVEAT: Towards Robust Computer-Use Agents in Incentive-Misaligned Environments # CAVEAT: Towards Robust Computer-Use Agents in Incentive-Misaligned Environments Yuxuan Li † † thanks: Work done during an internship at Microsoft Research. Affiliation: Carnegie Mellon University Email: [email protected] Will Epperson Affiliation: Microsoft Research Email: [email protected] Wesley…

  4. An unexamined cause of the OpenAI Hugging Face hacking incident: its binary performance metric — LessWrong — LessWrong

    # An unexamined cause of the OpenAI Hugging Face hacking incident: its binary performance metric — LessWrong Published: 2026-09-23T00:27:16+00:00 Source: lesswrong.com (lesswrong.com) Language: en ## Story # An unexamined cause of the OpenAI Hugging Face hacking incident: its binary performance metric * By [W Bradley Knox](/users/w-bradley-knox), [Serena Booth](/users/serena-booth), [Brian…

  5. OpenAI's Safety Evals Shrink Between Promise and Delivery - DEV Community — DEV Community

    # OpenAI's Safety Evals Shrink Between Promise and Delivery - DEV Community Published: 2026-09-23T08:18:20+00:00 Source: dev.to (dev.to) Language: en ## Story OpenAI's Safety Evals Shrink Between Promise and Delivery - DEV Community Peremptory Posted on Sep 23 • Originally published at peremptory.ai # OpenAI's Safety Evals Shrink Between Promise and Delivery Ten days ago, Sam Altman…

Daily TopFive for AI Safety Daily. Links above go to the original articles. Follow and rate AI Safety Daily on Apple Podcasts. Feedback? Email [email protected].

Tous les épisodes de AI Safety Daily

Paramètres

Affichage

Langue

Détermine les stations affichées en premier sur la page d’accueil.

Historique

Les stations et épisodes écoutés récemment et tes dernières recherches sont seulement enregistrés dans ce navigateur.

Notifications

Un court message quand un podcast que tu suis publie un nouvel épisode. Désactive un podcast précis avec la cloche sur sa page.

Compte

Enregistrez vos favoris – dans l’app et sur le site.

Raccourcis clavier

/
Ouvrir la recherche
Espace
Lecture ou pause
← →
Dans le lecteur : 15 s en arrière, 30 s en avant (épisodes) · dans une rangée : tuile suivante
↑ ↓
Dans une rangée : cible dans la tuile
Début / Fin
Dans une rangée : première ou dernière tuile
Esc
Fermer la fenêtre ou le menu
?
Cet aperçu
En cours de lecture

Voir la station

À suivre

    Ajoutez des épisodes avec « À suivre » – ils seront lus après l'épisode en cours.