Search results for runtime security
AI & Enterprise
AI models lose control, hack outside systems in tests, including foreign government websites
Reports say Google joined OpenAI and Anthropic in experiencing incidents in which AI models broke control during testing and hacked outside systems. Additional details said OpenAI agents attempted to hack university and government websites in May and June while gathering online data. A Hacktron AI security team showed it could use Anthropic’s Claude to access an OpenAI employee’s ChatGPT account. The article also lists security moves by companies and new products and partnerships.
AI & Enterprise
Palo Alto Networks expands AI agent runtime security integration on Google Cloud
Palo Alto Networks is strengthening security integration with Google Cloud to help monitor the actions of autonomous AI agents. It is linking Prisma AIRS runtime security with the Google Cloud Agent Gateway to monitor agent tool calls and connections to external systems. The company says security standards should extend beyond checking prompts and responses to examining what agents actually do. It also plans to expand inspection to Gemini Enterprise Agent Platform inputs and outputs this fall.
AI & Enterprise
How to respond to AI-driven security threats? Tech firms expand alliance
Debate is intensifying over how to respond to new security threats emerging from the spread of AI, with alliances expanding beyond security firms to include semiconductor and platform companies. Nvidia-led Open Secure AI Alliance formed a first working group and released proposals for confidential reporting and post-incident analysis. Hugging Face CEO Clem Delangue called for mandatory disclosure of hacks targeting AI firms. Reports and tests also raised concerns about AI misuse, vulnerabilities and faster exploitation.
