Tag: agents
-

OpenAI’s Codex Guide Is About Sticking With the Work
OpenAI put out a new Codex guide, and the point is not flash. It is continuity. The guide treats Codex like a persistent workspace that keeps context, handles longer jobs, and stays useful when work stretches past one prompt. That is a better signal than another grand claim about AI replacing everyone. Real work is…
-

New Attack Class Hijacks AI Coding Agents
Tenet Security disclosed Agentjacking, a novel attack that tricks AI coding agents into running malicious code using nothing but a fake error report. Here is how it works. Attackers craft error messages in Sentry, the error-tracking tool that every developer uses. These messages look legitimate. Agents like Claude Code, Cursor, and OpenAI Codex read the…
-

OpenAI Buys Ona: The Agent Era Needs Infrastructure
OpenAI just acquired Ona, a German startup that builds secure cloud environments for AI agents to do real work over hours and days, not minutes. This deal is about one thing: the bottleneck in AI is no longer raw inference speed. It is persistence. It is trust. When an AI agent needs to keep working…
-

AWS Just Gave AI Agents a Shared Memory Layer Now
AWS just pushed the idea of agent memory a lot closer to product reality. Its new Context layer builds a knowledge graph from existing data and lets agents pull from it without manual recuration. That sounds abstract, but it solves a problem every serious AI team runs into. Agents are only as good as the…
-

Google DeepMind Funds AI Safety Research With $10 Million
Google DeepMind just announced a 10 million dollar funding call for researchers studying a problem most people haven’t thought about yet: what happens when millions of AI agents built by different companies all interact with each other. These agents will communicate, negotiate, and transact with one another. They’ll need to do all that safely and…
-

University of Toronto Just Proved AI Can Build Adaptive Worms
Researchers at the University of Toronto published a proof-of-concept on June 2 that should wake up every security team in the world: an AI-powered computer worm that reasons about its targets and adapts its attack strategy as it spreads. Not following a script. Reasoning. Learning. Self-replicating. In lab tests, the worm achieved a 73.8% exploitation…
-

AWS Just Weaponized Context for Agents
Amazon quietly launched AWS Context-a knowledge graph service built for AI agents that learns as agents use it. Not manually curated. Not static. Learning from every agent decision, every tool call, every interaction. This is infrastructure for the next generation of AI work. The premise is straightforward: agents need context to reason well. Today, enterprises…
-

Readout mistakes can mislead LLM tool selection
People who build “tool-using” AI systems often focus on whether the model can spot the right option. But a recent study suggests the bigger failure point can show up later—after the model has already looked at the tools. The paper examines a benchmark called BFCL. In the results, the correct tool ended up with the…
