NewsThe Hacker News
DeepSeek Harness Flaw Let AI Agents Disable Their Own File Sandbox Without Approval
More on DeepSeek
Other advisories
- HighCVE-2026-82033Deepseek-ai Deepseek-harness: fehlende Authentifizierung für kritische Funktion
- HighCVE-2026-82033Deepseek-ai Deepseek-harness: fehlende Authentifizierung für kritische Funktion
- MediumCVE-2026-120204Deepseek-ai DeepInfer: Pfadüberschreitung (Path Traversal)
- MediumCVE-2026-120204Deepseek-ai DeepInfer: Pfadüberschreitung (Path Traversal)
- MediumCVE-2026-115914vLLM DeepSeek-OCR image processing vulnerable to denial of service
- MediumCVE-2026-115914vLLM DeepSeek-OCR image processing vulnerable to denial of service
- Not ratedosv:CVE-2026-115914vLLM DeepSeek-OCR image processing vulnerable to denial of service
- HighCVE-2026-116368DeepSeek Harness: fehlerhafte Rechtevergabe
Other news entries
- 26,400 live cloud and AI keys found in public Gitee repositories, led by Alibaba Cloud, DeepSeek and OpenAIBreachpwn.watch
- China Defends DeepSeek After US Ban of Its AI From Government SystemsNewsSecurityWeek
- U.S. Orders DeepSeek Removed From Federal Civilian Systems, Bans New PurchasesNewsThe Hacker News
- US bans DeepSeek across federal agencies, putting AI supply chains under scrutinyNewsCSO Online
- DHS bans DeepSeek products across US federal civilian agenciesNewsBleepingComputer
- DeepSeek deep-sixed at all US federal civilian agenciesNewsThe Register — Security
- New ClosedQuorum Windows malware uses AI for attack decisionsNewsBleepingComputer
- Anthropic Says Seven China-Based AI Labs Ran Industrial-Scale Claude Distillation AttacksNewsThe Hacker News
Source: https://thehackernews.com/2026/09/deepseek-harness-flaw-let-ai-agents.html
ID