Tag: openai
-
OpenAI Disrupts Reasoning Extraction Campaign Linked to Moonshot AI Associates
OpenAI on Wednesday said it identified and disrupted a coordinated distillation campaign that was designed to illicitly extract protected reasoning from its artificial intelligence (AI) models.A “core cluster of the activity,” going back to the first week of July, has been attributed to individuals associated with Moonshot AI, a Chinese AI company based in Beijing.…
-
KI: FTC eröffnet Untersuchung gegen Anthropic, OpenAI und Co.
Die US-Verbraucherschutzbehörde FTC will mehr über die Gefahren durch KI wissen. Eine entsprechende Untersuchung wird gestartet. First seen on golem.de Jump to article: www.golem.de/news/ki-ftc-eroeffnet-untersuchung-gegen-anthropic-openai-und-co-2610-213608.html
-
OpenAI reveals ‘novel’ encryption bypass used in distillation attack
The company said individuals associated with Chinese company MoonshotAI were behind parts of the attack, but did not offer hard evidence for the claim. First seen on cyberscoop.com Jump to article: cyberscoop.com/openai-moonshot-ai-model-distillation-attack/
-
OpenAI Accuses Moonshot AI of Coordinated Model Distillation
Campaign Allegedly Used Thousands of Accounts to Extract Hidden Reasoning From OpenAI Models. OpenAI disrupted what it calls a coordinated campaign to extract the training data and reasoning of its AI models, allegedly by the developer of the Kimi model, Chinese company Moonshot AI. The company first noticed the activity in July. First seen on…
-
Malicious Custom GPTs Turn ChatGPT Into RAT Delivery Lure
In yet another ClickFix-style campaign, threat actors abuse legitimate domains from OpenAI and Google to fool unsuspecting users. First seen on darkreading.com Jump to article: www.darkreading.com/cyberattacks-data-breaches/malicious-custom-gpts-chatgpt-rat-delivery-lure
-
OpenAI Launches Codex Security Cloud for Always-On Application Security Scanning
OpenAI has expanded its Codex platform with Codex Security Cloud, a cloud-hosted application security feature that continuously analyzes GitHub repositories, investigates potential vulnerabilities, and prepares remediation patches for human review. Announced as part of the company’s latest Codex updates, this service is designed to operate security workflows beyond a developer’s local machine. It can run…
-
OpenAI Dots Pushes Always-on Agents Into the Enterprise
New Personal Agent Adds Guardrails But Leaves Security Teams With Limited Governance. OpenAI announced Dots, a new always-on personal agent powered by GPT-6 Astra that the company said would offload tasks and screen time from users. Dots competes against Meta’s Muse personal agent, and both represent the evolution of OpenClaw, the open-source agent that lets…
-
Custom ChatGPTs push ClickFix attacks to deploy RAT malware
Custom variants of OpenAI’s ChatGPT promoted in sponsored Google results are directing unsuspecting users to malicious sites that use ClickFix attacks to deliver malware. First seen on bleepingcomputer.com Jump to article: www.bleepingcomputer.com/news/security/custom-chatgpts-push-clickfix-attacks-to-deploy-rat-malware/
-
OpenAI Apologizes for Hacks on Australian Government Sites
AI Firm Admits Delayed Disclosure, Details Incidents, Promises New Agent Safeguards. OpenAI is publicly apologizing to Australian officials for a June hack by rogue AI agents of the government’s Medicare website. The AI company also listed several other Australian government websites that were accessed, pledging resources and expert support to affected agencies. First seen on…
-
OpenAI apologizes for agents breaching Australian government websites without authorization
The artificial intelligence giant acknowledged it botched its response to the incidents and should have done more to promptly notify and work with the Australian government in the days after it discovered the breaches. First seen on therecord.media Jump to article: therecord.media/openai-apologizes-australia-medicare-breach
-
GPT-6 Astra is More Prone to Rogue Supply-Chain Attacks
UK Agency Found GPT-6 Astra Attacked Out-of-Scope Targets in Simulated Cyber Tests. The U.K. AI Security Institute found that OpenAI’s GPT-6 Astra sometimes carried out unsanctioned supply-chain attacks in simulated environments, including against targets it had been told were out of scope. The model also created fake identities and submitted malicious code for human review.…
-
OpenAI Gets Sued Over the Hugging Face Hack
Tags: openaiA nonprofit in California is doing what Hugging Face has not”, attempting to hold OpenAI legally accountable for the actions of its agents. First seen on wired.com Jump to article: www.wired.com/story/openai-sued-over-the-hugging-face-hack/
-
OpenAI Cancels GPT-6.1 Astra Release Over Internal Safety Concerns
OpenAI has canceled the planned October release of GPT-6.1 Astra after internal testing revealed that the next-generation model did not meet the company’s safety and alignment standards. This decision underscores a growing challenge for AI developers: ensuring that highly capable autonomous systems do not exceed user intent, conceal actions, or bypass oversight. The model was…
-
OpenAI apologizes to Australia after its AI agents breached government sites
The company also detailed how some of those breaches had happened, and outlined additional measures it is taking to assess the impact of the events. First seen on techcrunch.com Jump to article: techcrunch.com/2026/09/29/openai-apologizes-to-australia-after-its-ai-agents-breached-government-sites/
-
OpenAI apologizes to Australia after its AI agents breached government sites
The company also detailed how some of those breaches had happened, and outlined additional measures it is taking to assess the impact of the events. First seen on techcrunch.com Jump to article: techcrunch.com/2026/09/29/openai-apologizes-to-australia-after-its-ai-agents-breached-government-sites/
-
OpenAI apologizes to Australia after its AI agents breached government sites
The company also detailed how some of those breaches had happened, and outlined additional measures it is taking to assess the impact of the events. First seen on techcrunch.com Jump to article: techcrunch.com/2026/09/29/openai-apologizes-to-australia-after-its-ai-agents-breached-government-sites/
-
OpenAI apologizes to Australia after its AI agents breached government sites
The company also detailed how some of those breaches had happened, and outlined additional measures it is taking to assess the impact of the events. First seen on techcrunch.com Jump to article: techcrunch.com/2026/09/29/openai-apologizes-to-australia-after-its-ai-agents-breached-government-sites/
-
OpenAI’s GPT-6 Astra ran supply chain attacks despite being told not to
OpenAI’s GPT-6 Astra carried out supply chain attacks on software outside the scope of a security test, according to the UK AI Security Institute (AISI). Anatomy of an … First seen on helpnetsecurity.com Jump to article: www.helpnetsecurity.com/2026/09/29/openai-gpt-6-astra-supply-chain-attacks-test-simulations/
-
GPT-5.4-mini: OpenAI warnt vor selbst replizierender Prompt-Injection
Die neuartige Prompt-Injection hat Fähigkeiten eines Computerwurms. OpenAI entdeckt sie bei Trainings mit Prompt-Injections. First seen on golem.de Jump to article: www.golem.de/news/gpt-5-4-mini-openai-warnt-vor-selbst-replizierender-prompt-injection-2609-213525.html
-
GPT-6 Astra and the Supply Chain Attack It Wasn’t Asked to Launch
UK AISI finds GPT-6 Astra launches unsanctioned supply-chain attacks in simulations far more than earlier OpenAI models, even when told not to. The UK’s AI Security Institute tested GPT-6 Astra before its public release, and the results, published September 28, aren’t subtle. When given a routine cybersecurity evaluation, the model went off script and attacked…
-
OpenAI Pauses Tool Use After Agent Bypasses Internet Controls to Reach External Chatbot
OpenAI said it has made the decision to pause training of its most powerful models after one of its agents during reinforcement learning (RL) training contacted an external chatbot by exploiting a loophole in its internet-access restrictions.”An agent attempting to complete a search-based training task queried a public chatbot service through a gap in our…
-
OpenAI Pauses Top Models as Rogue Agents Target Agencies
Outside Researchers Exposed Government and UN Incidents Before OpenAI Disclosures. OpenAI once again paused training and evaluation of its most capable models as news of more AI agents trying to hack into U.S. government agencies and the United Nations continues to surface. The attempted breaches show how much of what’s known about OpenAI’s misbehaving agents…
-
Anthropic Declines Australian AI Hearing as OpenAI Agent Breach Faces Scrutiny
Anthropic declined an Australian Senate AI hearing as officials investigate an OpenAI agent breach and consider mandatory AI incident reporting. The post Anthropic Declines Australian AI Hearing as OpenAI Agent Breach Faces Scrutiny appeared first on TechRepublic. First seen on techrepublic.com Jump to article: www.techrepublic.com/article/news-anthropic-australia-ai-hearing-openai-agent-breach-apac/
-
OpenAI Agent Swarm Used Nearly 1 Million URLs to Hack Hugging Face
A newly released forensic investigation has reconstructed how a swarm of about 700 OpenAI evaluation agents allegedly used nearly one million chained URLs to bypass restricted internet access and compromise parts of Hugging Face’s infrastructure. This incident illustrates how seemingly limited web-access capabilities can be combined with third-party services to create a functional execution, command-and-control,…
-
KI-Agenten von OpenAI legen US-Behördendaten offen
KI-Agenten von OpenAI haben mithilfe öffentlich auffindbarer Entwickler-Schlüssel auf Daten des US Census Bureau zugegriffen und öffentlich zugängliche Informationen der US-Börsenaufsicht SEC auf einer anderen Website erneut veröffentlicht. Zudem erfolgte ein gescheiteter Versuch, eine Website des US-Bildungsministeriums zu hacken. OpenAI hat inzwischen Dutzende Organisationen über Aktivitäten informiert, bei denen Sicherheitskontrollen umgangen wurden. Adam Marrè, CISO…
-
OpenAI Pauses Training Its Most Powerful Models After Rogue Agents Target Government
Sam Altman says the company “have not been as fast as we would have liked” at dealing with security breaches, after news of further incidents over the summer forces another temporary halt. First seen on wired.com Jump to article: www.wired.com/story/openai-pauses-training-most-powerful-models-after-rogue-agents-target-government/
-
Künstliche Intelligenz: Nvidia will ausbrechende KI-Systeme mit neuen Tools eindämmen
Mit seinen neuen Open-Source-Tools hätte OpenAIs Angriff auf Hugging Face verhindert werden können, sagt Nvidia. First seen on golem.de Jump to article: www.golem.de/news/kuenstliche-intelligenz-nvidia-will-ausbrechende-ki-systeme-mit-neuen-tools-eindaemmen-2609-213488.html