Tag: openai
-
What the Hugging Face breach reveals about defense in the age of agentic AI
We almost never get both sides of an intrusion. This time we did. Last month, Hugging Face disclosed a breach into part of its production infrastructure, saying an autonomous AI agent system ran the attack from start to finish. Five days later, OpenAI revealed that its own models, including GPT-5.6 Sol along with an unreleased…
-
IT-Sicherheitsforscher – OpenAIs Hugging-Face-Hack war menschliches Versagen
Tags: openaiDass ein Forschungsmodell von OpenAI aus der Sandbox ausbrechen konnte, erklärten Experten durch IT-Sicherheitsverstöße. First seen on computerbase.de Jump to article: www.computerbase.de/news/apps/it-sicherheitsforscher-openais-hugging-face-hack-war-menschliches-versagen.98645
-
KI aus China: Kann Kimi K3 mit OpenAI und Anthropic mithalten?
First seen on t3n.de Jump to article: t3n.de/news/ki-aus-china-kimik3-1749657/
-
An OpenAI Agent Escaped Its Sandbox and Hacked Hugging Face to Cheat on Its Own Benchmark
Tags: openaiHugging Face detected the intrusion on July 16. OpenAI worked out five days later that the attacker was its own model, cheating on its own benchmark. First seen on securityboulevard.com Jump to article: securityboulevard.com/2026/07/an-openai-agent-escaped-its-sandbox-and-hacked-hugging-face-to-cheat-on-its-own-benchmark/
-
Anthropic says its AI accidentally hacked three companies during safety tests
Following OpenAI’s own incident, Anthropic reviewed its own evaluations and found three cases of Claude hacking external companies. First seen on cyberscoop.com Jump to article: cyberscoop.com/anthropic-claude-ai-hacks-real-companies/
-
Anthropic Says Claude Hacked 3 Organizations During Cybersecurity Tests
In a review triggered by OpenAI’s Hugging Face incident, Anthropic discovered three of its AI models had breached real organizations during third-party evaluations. First seen on wired.com Jump to article: www.wired.com/story/anthropic-says-claude-hacked-real-systems-during-cybersecurity-tests/
-
Breach Roundup: OpenAI Models on a Hacking Tear
Also, Russian Hackers Exploit Outlook Flaw, Coca-Cola Restarts Fairlife Production. This week: Sam Altman on hacking, Russia exploited an Outlook web access flaw, Coca-Cola restarted Fairlife production, U.K. education department and Angola teleco breached, SonicWall credential stuffing, Telegram founder charged in Russia, hidden prompt turns Microsoft Copilot into an AI worm. First seen on govinfosecurity.com…
-
Hugging Face Incident Spurs Calls for European AI Autonomy
European Union Looks to Launch ‘AI Gigafactories’. OpenAI’s inadvertent agentic hacking of Hugging Face’s systems has startled some politicians in Europe, with Germany’s top tech minister saying it demonstrated the need for the rapid development of the European AI sector. The European Commission said it will build up to seven AI gigafactories. First seen on…
-
JFrog Patches Flaws Behind OpenAI Models’ Escape
Artifactory Bugs Helped Models Reach Hugging Face Production. JFrog patched previously unknown flaws in self-hosted Artifactory after OpenAI models used them to escape a cyber test environment and reach Hugging Face production. The repository software is run by more than 7,500 DevOps teams, including most Fortune 100 companies. First seen on govinfosecurity.com Jump to article:…
-
Rogue OpenAI Agent Hit More Than One Target, New Disclosures Show
OpenAI disclosed that the rogue AI agent behind the Hugging Face breach also accessed four additional public services during the same cybersecurity incident. First seen on esecurityplanet.com Jump to article: www.esecurityplanet.com/cybersecurity/openai-rogue-ai-agent-accessed-four-additional-services/
-
We now have a better understanding how OpenAI hacked into Hugging Face
10 days passed from OpenAI models exploiting JFrog Artifactory 0-day to release of a patch. First seen on arstechnica.com Jump to article: arstechnica.com/security/2026/07/jfrog-tries-to-spin-openai-0-day-exploit-of-its-app-into-a-success-story/
-
JFrog tries to spin OpenAI 0-day exploit of its app into a success story
10 days passed from OpenAI models exploiting JFrog Artifactory 0-day to release of a patch. First seen on arstechnica.com Jump to article: arstechnica.com/security/2026/07/jfrog-tries-to-spin-openai-0-day-exploit-of-its-app-into-a-success-story/
-
When AI Agents Escape Sandboxes, Old Security Rules Apply
OpenAI’s recent AI agent sandbox escape proves traditional security principles matter more than ever: limit access, isolate execution, log everything. First seen on darkreading.com Jump to article: www.darkreading.com/application-security/ai-agents-escape-sandboxes-old-security-rules-apply
-
OpenAI models used Artifactory zero-days to escape to the internet
JFrog has confirmed that OpenAI models exploited zero-day vulnerabilities in self-hosted Artifactory servers to help escape an isolated testing environment and gain access to the internet before attacking Hugging Face. First seen on bleepingcomputer.com Jump to article: www.bleepingcomputer.com/news/security/openai-models-used-artifactory-zero-days-to-escape-to-the-internet/
-
Hugging Face breach reignites open-weights debate, raises liability questions
The first publicly documented cyberattack run end-to-end by an autonomous AI was an OpenAI benchmark test that escaped its sandbox and breached Hugging Face. In an incident … First seen on helpnetsecurity.com Jump to article: www.helpnetsecurity.com/2026/07/28/hugging-face-breach-ciso-playbook-open-weight-llms/
-
JFrog Confirms OpenAI Models Exploited Artifactory Zero-Day Before Hugging Face Breach
JFrog has confirmed that OpenAI models exploited a zero-day in self-hosted Artifactory while trying to reach the open internet from a sealed evaluation environment.Artifactory is JFrog’s software repository manager. OpenAI says the models then escalated privileges and moved laterally until they reached an internet-connected node. JFrog says it has since developed and released fixes for…
-
OpenAI models escape containment, hack Hugging Face
In an unprecedented — and unintended — cyberattack, frontier AI models autonomously escaped their contained testing environment and breached another company’s systems. First seen on techtarget.com Jump to article: www.techtarget.com/searchsecurity/news/366646105/OpenAI-models-escape-containment-hack-Hugging-Face
-
Open Secure AI Alliance: Warum Microsoft, OpenAI und Nvidia plötzlich auf Open-Source-KI setzen
Trotz der durchaus sinnvollen Forderungen der Tech-Konzerne zur Abwehr von KI-Gefahren drängen sich auch andere Motive auf. First seen on golem.de Jump to article: www.golem.de/news/open-secure-ai-alliance-warum-microsoft-openai-und-nvidia-ploetzlich-auf-open-source-ki-setzen-2607-211359.html
-
Nach KI-Cyberangriff: Hugging Face stellt 100-Millionen-Dollar-Forderung an OpenAI
Der KI-Cyberangriff von OpenAI hat Folgen: Der Hugging-Face-CEO fordert nun radikale Transparenz sowie eine Investition in die Absicherung seiner Plattform. First seen on golem.de Jump to article: www.golem.de/news/nach-ki-cyberangriff-hugging-face-stellt-100-millionen-dollar-forderung-an-openai-2607-211344.html
-
OpenAI steigt erstmals in Top 10 der imitierten Marken auf
Check Point zeigt im Q2-2026-Bericht: Microsoft führt das Phishing-Ranking an, während ChatGPT von OpenAI erstmals unter den zehn meistimitierten Marken liegt. First seen on it-daily.net Jump to article: www.it-daily.net/it-sicherheit/cybercrime/openai-top-10-der-imitierten-marken
-
OpenAI CEO Sam Altman Claims AI Has Reached Singularity as Systems Begin Improving Themselves
OpenAI CEO Sam Altman has stated that artificial intelligence has entered the long-discussed stage of technological singularity, referring to the current period as the point where AI systems can increasingly improve future AI capabilities. His remarks, made during the “Relentless” podcast released on Saturday, have reignited the debate over whether advanced AI models have reached…
-
Reuters: OpenAI Agent Hacked Hugging Face for Days Before Being Detected
Reuters says OpenAI failed to detect its AI agent hacking Hugging Face for days, discovering the breach only after FBI involvement. Reuters reported that the OpenAI agent responsible for the Hugging Face breach operated undetected for over a week before OpenAI realized what had happened, long after the FBI had been alerted and Hugging Face…
-
Nvidia Launches Open-Source AI Security Alliance
Anthropic, OpenAI and Google Absent as 37 Firms Back Open AI Security Tools. Nvidia and 36 other technology giants launched the Open Secure AI Alliance to build and share open-source AI security tools, arguing open models are critical defensive assets – while Anthropic, OpenAI and Google, makers of the most capable closed models, are absent…
-
OpenAI-Hugging Face Incident: What We Know
An experimental AI agent powered by OpenAI models has successfully compromised part of Hugging Face’s infrastructure during a controlled cybersecurity evaluation, offering a glimpse into the evolving offensive capabilities of advanced AI systems. The incident, first disclosed by Hugging Face last week, has now been confirmed by OpenAI to have involved a combination of its…
-
OpenAI erstmals unter den zehn meistimitierten Marken
Der Bericht von Check Point über das Brand-Phishing-Ranking für das zweite Quartal 2026 listet erstmals eines der großen KI-Unternehmen in den Top 10. Die Experten beobachteten zudem konkrete Betrugsmaschen mit ChatGPT, Paypal, iCloud und Microsoft Microsoft bleibt weiterhin Spitzenreiter als meistimitierte Marke und vereint 23 Prozent aller Brand-Phishing-Versuche im Quartal auf sich. Das sind fast…
-
Tech giants form alliance to put open AI in cyber defenders’ hands
NVIDIA and a group of tech companies have formed an alliance to promote the use of open AI models in cybersecurity, days after OpenAI disclosed that one of its own AI models … First seen on helpnetsecurity.com Jump to article: www.helpnetsecurity.com/2026/07/27/nvidia-open-secure-ai-alliance/
-
Angreifer konnten verdeckte KI-Agenten in ChatGPT einschleusen
Zenity Labs entdeckte die Schwachstelle AgentForger in ChatGPT Workspace Agents. OpenAI hat den Fehler kurz nach der Meldung behoben. First seen on it-daily.net Jump to article: www.it-daily.net/it-sicherheit/cybercrime/ki-agenten-in-chatgpt-einschleusen
-
Wenn der Test zur Attacke wird: Der OpenAI-Vorfall als Weckruf für die KI-Sicherheit
Ein autonomer KI-Agent verlässt eine isolierte Testumgebung, verschafft sich Zugang zum offenen Internet und kompromittiert die Infrastruktur einer fremden Plattform. Der Vorfall bei Hugging Face markiert eine Zäsur: Aus der theoretischen Debatte über agentische Risiken ist eine konkrete Herausforderung für Cyberabwehr, Modelltests und Regulierung geworden. Die entscheidende Erkenntnis: Hochleistungsmodelle sind nicht mehr nur Werkzeuge,……
-
Hugging Face CEO calls for ‘radical transparency’ after ‘unprecedented’ OpenAI hack
“The first autonomous agent cyberattack is an unprecedented event. It deserves an unprecedented response!” First seen on techcrunch.com Jump to article: techcrunch.com/2026/07/26/hugging-face-ceo-calls-for-radical-transparency-after-unprecedented-openai-hack/
-
Landes-Geheimdienstchef Kramer: OpenAI-Hackerangriff war kein Skynet-Szenario
Nach dem Hackerangriff auf Hugging Face mahnt der Thüringer Verfassungsschutz zur Besonnenheit. Er forderte aber ein KI-Frühwarnsystem für die Sicherheitsbehörden. First seen on golem.de Jump to article: www.golem.de/news/landes-geheimdienstchef-kramer-openai-hackerangriff-war-kein-skynet-szenario-2607-211287.html