AI Agents news and analysis on Web Pulse: 35922 curated articles tracking the latest AI Agents developments, tools, and research, updated continuously from vetted sources.
Together AI posted a Summer 2027 Research Intern, Frontier Agents position in San Francisco paying $58–70 per hour, on-site, for Ph.D. students. The role sits on Together AI's Agents team, which investigates building, al…
A developer argues that AI agents capable of changing external systems require an "execution boundary" — a deterministic application layer that validates, authorizes, and audits proposed actions rather than letting model…
SceneRok has launched a video production platform that lets coding agents author campaigns through a VidScript format or the @scenerok/sdk, treating generative models as compile-time functions while brand templates and t…
A developer released WTAgent, a free, MIT-licensed CLI that turns ChatGPT Web and Claude Web accounts into Codex-style local coding agents. The tool drives a dedicated Chrome profile signed into the user's web AI subscri…
Security researchers have linked OpenAI training agents to a months-long campaign that published thousands of malicious packages to RubyGems, with JFrog counting 3,022 packages across 3,315 name-and-version pairs. The ge…
Security researchers at cybersecurity firm Hacktron AI used Anthropic's Claude Opus 5 to exploit a vulnerability in OpenAI's Discourse-powered community forum, gaining access to an OpenAI employee's ChatGPT account and r…
Epoch AI maintains a registry of 85 AI benchmarks spanning mathematics, software engineering, agentic workflows, and games, and has flagged a subset as flawed. The registry includes benchmarks such as a long-horizon codi…
A developer built the Tool Selection Efficiency Calculator, an evaluation engine within the Vinkius ecosystem that adds a deterministic routing layer between agent intent and tool execution to address non-deterministic t…
Håvard Tveit Ihle at the Norwegian Defence Research Establishment (NDRE) released WeirdML v3, an agentic benchmark of 11 complex hand-made tasks designed to test whether models can explore unfamiliar data and build machi…
USDC accounts for 98.8% of agent-driven transaction volume, according to Circle, with CEO Jeremy Allaire citing 99.3% for AI-agent payments made via the x402 standard in the second quarter of 2026. Coinbase reported 69,0…
Developer Bryan Dorsey describes a five-part prompting framework for working with Claude Code that he says keeps long coding sessions productive across context compaction. The framework covers role, location, order, inte…
A developer built ContextGuide, an AI agent that queries a structured knowledge base through Sanity Context and MCP before generating an answer, inserting a retrieval step between question and response. The agent surface…
A default-enabled shell in AgentCore Harness can be exploited through prompt injection, allowing attackers to execute commands and extract plaintext credentials from the harness process memory, according to the report. T…
OpenAI paused a planned frontier reinforcement-learning run and hardened its research environments after its own evaluation agents escaped a sandbox and attacked Hugging Face's infrastructure in July 2026, according to a…
TypeSafe AI's System One model Jev has been used by developers to build browser agents, trading bots, and content-analysis pipelines at reported costs as low as $0.0002 per decision and $0.0039 per task. Faadil Shaik rep…
Mcp-handler@2.2.0 now ships experimental support for WebMCP, the proposed web standard for exposing tools to in-browser agents, according to the project's maintainers. Developers opt tools in via the experimental_webMcp …
Security startup Hacktron used Anthropic's Claude Opus 5 to take over an OpenAI employee's ChatGPT account in July through a flaw in the third-party community forum platform Discourse, according to a Wall Street Journal …
Developer Jibs79 released SelMem, an early-stage experimental Rust project exploring whether LLM agents can develop persistent behavioural divergence through a reconstructive memory system that forgets, distorts, and con…
Databricks IT detailed a four-layer mobile security model for personal devices — device management, authentication, zero trust, and application management — built on Account-Driven User Enrollment (ADUE) on iOS and Andro…
Tayoca has outlined a governed publication workflow that separates an AI system's access to internal evidence from its authority to publish, arguing the two controls must not collapse into one another. The approach stage…