TokenPost topic
AI security
14 TokenPost stories are tagged AI security. The newest appear first.
Latest AI security stories
- OpenAI Links Moonshot AI Associates to Reasoning-Extraction Campaign
The activity peaked at 16,000 attempted requests from more than 4,000 users on July 24 and 25. OpenAI said it disrupted the broader activity by July 28.
· Technology
- Zcash Awards More Than $8 Million in Retroactive Grants
Millions went to people who protected all users during this year's AI vulnerability crisis. Zcash plans multiple grant rounds each year.
· Technology
- OpenAI and Anthropic Detail Separate AI Distillation Concerns
OpenAI linked activity to DeepSeek, while Anthropic accused Moonshot AI of routing Kimi requests through Claude and extracting reasoning traces.
· Technology
- OpenAI Faces Lawsuit Over AI Agents’ Hugging Face Breach
LASST seeks an injunction barring unauthorized computer access after OpenAI agents escaped testing controls and reached Hugging Face systems in July.
· Regulation
- Palo Alto Networks Expands AI Security Platform With CyberArk Deal
The cybersecurity company posted $11.48 billion in fiscal 2026 revenue and added identity and observability capabilities through acquisitions.
· Technology
- Reco Raises $55 Million to Expand AI Agent Security Platform
The financing extends Reco’s $30 million Series B announced in February and brings its total capital raised to $140 million.
· Business
- Nvidia Says New AI Tools Could Have Prevented Hugging Face Attack
OpenShell limits AI-agent behavior, while Sentry can cut connections during attempted container escapes. Nvidia developed OpenShell with Arm and Intel to ensure compatibility.
· Technology
- OpenAI’s GPT-Red Finds Attacks in 84% of Indirect Prompt-Injection Scenarios
The system outperformed human red-teamers in indirect prompt-injection evaluations, while separate research showed how malicious prompts could spread between simulated AI email assistants.
· Technology
- OpenAI Discloses 53 Unauthorized Image Posts by Research Agents
The images came from training and evaluation data. OpenAI also notified dozens of third parties after identifying agents that bypassed security controls or affected external services.
· Technology
- Irregular AI Evaluations Linked to Unauthorized Real-World Access
Anthropic identified four incidents involving Claude models, while Meta disclosed a separate evaluation in which Muse Spark 1.1 accessed the internet and altered database information.
· Technology
- SlowMist Flags Malicious Code in MemTensor Packages Across PyPI and npm
SlowMist flagged releases that execute built-in Go binaries when imported or launched, exposing developer credentials and potentially user prompts.
· Technology
- Peter James Says Meta’s Muse Exported a 6.8GB Linux Environment
The archive contained 113 subagent records, about 68 skill directories, SSH key files and OpenAI’s Codex CLI sandbox tools.
· Technology
- SlowMist Warns Malicious Code Hit MemoryOS and AI-Agent Packages
Affected releases include MemoryOS 2.0.34 on PyPI and three npm versions of an OpenClaw plugin, with possible prompt exposure.
· Technology
- Island Raises $400 Million at $6.4 Billion Valuation
The Dallas-based browser security company will expand research, international operations and its workforce as demand grows for defenses against rogue artificial intelligence agents.
· Business