Unlocking AI's Potential: Key Advances in Agent Security and Real-Time Interactions

In a series of groundbreaking developments, leading tech companies and startups are pushing the boundaries of artificial intelligence (AI) with new advancements in agent security and real-time interactions. OpenAI acquires Ona to enhance long-running agent capabilities, while Google launches a startup school focused on agentic AI. Additionally, Kaggle introduces a competition to test the security of AI agents, and Kimi.ai open-sources a more efficient coding model.

OpenAI Acquires Ona for Long-Running Agents

OpenAI announces the acquisition of Ona, a company specializing in secure cloud execution and orchestration. This move aims to integrate Ona's technology into the Codex platform, enabling persistent, customer-controlled environments where AI agents can work across extended periods and sessions.

Kaggle Competition Focuses on AI Agent Security

Kaggle, in partnership with OpenAI, Google, and IEEE, launches a new competition: AI Agent Security: Multi-Step Tool Attacks. The competition challenges participants to develop attack algorithms that stress-test tool-using AI agents. The goal is to identify multi-step attack paths and make failure modes in these systems more reproducible, measurable, and better understood.

Google Startup School: Agentic AI

Google kicks off its Startup School series with a focus on agentic AI. The introductory session explores the multimodal capabilities of the Gemini Live API, demonstrating ultra-low latency voice and vision interactions. Participants will learn prompt design for natural, interruptible conversations and how to transition from rapid prototypes to stateful applications using Firebase Auth and Firestore.

Kimi.ai Open-Sources New Coding Model

Moonshot AI releases Kimi-K2.7-Code, an open-source coding model that uses 30% fewer reasoning tokens than its predecessor. This new model is faster, smarter, and cheaper to run, making it a significant improvement over K2.6. Developers can access it via the Kimi API, through the Kimi Code CLI agent, or by self-hosting the model.

Industry Context and Implications

These advancements reflect the rapid evolution of AI and its growing importance in various industries. By enhancing agent security and real-time interactions, companies aim to build more reliable and efficient AI systems. These developments also highlight the need for robust security measures as AI agents take on more complex and critical tasks.

References

• AI News Briefs BULLETIN BOARD for June 2026 | Radical Data Science