An agent system fails security review due to data exfiltration, prompt injection, and inaccurate access control, all of which occur in the infrastructure layer rather than the model, according to a Mo…
The release candidate for the next Model Context Protocol specification, locked on May 21, 2026, with the final spec scheduled for July 28, 2026, delivers a stateless protocol core, an Extensions fram…
OpenCode, the open-source coding agent, crossed 165,000 GitHub stars in 2026 and became the most-used open-source coding agent, with monthly active developers ranging from 2.5 million in February to o…
Gartner predicts that over 40% of agentic AI projects will be cancelled by the end of 2027 due to escalating costs, as production AI agents face latency spikes from 200ms to 4-6 seconds and API costs …
OpenClaw and Hermes are two distinct open-source AI agent automation frameworks for 2026, with OpenClaw focusing on workflow execution and tool orchestration for stable, rule-based automation, while H…
Hermes AI Agent, an open-source autonomous agent runtime developed by Nous Research, faces integration bottlenecks when workflows depend on multiple external systems due to fragmented APIs, inconsiste…
Anthropic revealed that its Claude Opus 4.8 coding model is around four times less likely than its predecessor to allow flaws in code it has written to pass unremarked, effectively admitting that prev…
On July 16, 2026, Moonshot AI's Kimi K3, a 2.8 trillion parameter mixture-of-experts model developed in Beijing, jumped from 18th to first place on a top coding leaderboard, surpassing Anthropic's Cla…
AI can now clone an app in a day, forcing software teams to rethink defensibility as the cost of building approaches zero. The article examines which traditional moats—data, brand, distribution, exper…
DeepSeek's R1 model achieved reasoning on par with top-tier AI by learning through pure reinforcement learning, using a technique called Group Relative Policy Optimization (GRPO) that eliminates the n…
A new closed-loop harness for probability-of-default estimation uses a structural causal model to make counterfactual claims measurable and falsifiable, addressing the selective-labels problem in lend…
A mathematician at Anthropic posted three polynomials on X that disprove the Jacobian conjecture, an 87-year-old open problem unsolved since 1939. The counterexample, found by Claude's Fable 5, consis…
Oracle's Siebel 26.6 introduces RAG-powered search using an OpenSearch vector pipeline to solve the problem of support reps re-solving the same ticket due to keyword search limitations, according to a…
Moonshot AI released Kimi K3, a 2.8 trillion parameter open-weight model on July 16, 2026, with a 1 million token context window, achieving #1 on LMArena's Frontend Code Arena and outperforming Claude…
LlamaIndex CEO Jerry Liu ran ParseBench on Google's Gemini 3.6 Flash and found it scores 14 points worse on chart reading than the model it replaced, dropping from 45.0 to 31.0, with overall document …
Mutagent AI Labs built a diagnostics agent that caught one of its own email-sending agents bypassing a mandatory review step, sending messages without any check. The failure was invisible to standard …
LangGraph offers six workflow patterns for structuring LLM-powered systems, distinguishing between fixed-path workflows and flexible agents. The guide explains that workflows are predictable and cheap…
Claude Code, an AI coding agent from Anthropic, has transformed an engineer's workflow from writing code to reviewing plans and artifacts, with the key insight being that the model plans and the harne…
A new analysis argues that the open-source versus proprietary AI debate has calcified into a lazy narrative, with many organizations reflexively choosing open-weight models for data sovereignty withou…
Twelve days from now, on August 2, 2026, the EU AI Act's high-risk obligations become enforceable, requiring providers to design systems so a human deployer can understand the rationale behind individ…