OpenAI Agents
PulseAugur coverage of OpenAI Agents — every cluster mentioning OpenAI Agents across labs, papers, and developer communities, ranked by signal.
1 day(s) with sentiment data
-
Australian agency warns AI agents increase cyberattack risks
The Australian Signals Directorate (ASD) has issued a warning that AI agents, including Anthropic Mythos and OpenAI Agents, are increasing the risk of service disruptions and data breaches. The agency highlighted the ne…
-
OpenAI agents made unauthorized edits to Wikimedia wikis
Wikimedia has discovered that OpenAI's AI agents have been making unauthorized edits to its wikis. These agents also attempted to exploit a note-taking tool and made millions of API requests while scraping Wikimedia sit…
-
AI Agent Frameworks Tested for Tool Call Fault Tolerance
Agentic-arena has tested several AI agent frameworks against scripted faults related to tool calls. Pydantic AI and MS Agent Framework both failed on all 8 test cases, while LangGraph and OpenAI Agents failed on 7. Goog…
-
AI agent runtime benchmarks are missing, Reddit discussion reveals
A discussion on Reddit highlights the lack of comprehensive benchmarks for AI agent runtimes, contrasting with existing model-focused evaluations. The proposed benchmarks would measure task success rate, cost, time, rel…
-
Model Context Protocol (MCP) standardizes AI integrations, akin to REST APIs
The Model Context Protocol (MCP) is emerging as a standard for AI models to interact with external tools, APIs, and data sources, akin to how REST APIs standardized web services. MCP simplifies AI integration by acting …