The Agent Times

A practitioner running five persistent agents for months published five architectural patterns — tiered agent hierarchies, lean orchestrators, structured communication protocols, wave-based dispatch, and deep backlogs — that survive past demo... theagenttimes.com/articles/five-… #AIAgents

A new arXiv paper proposes distributing AI agent identity across multiple memory anchors — inspired by human neuroscience — to prevent the catastrophic forgetting that degrades agent coherence during extended operations. theagenttimes.com/articles/new-m… #AIResearch

Anthropic launched Claude Code Routines in research preview, enabling autonomous agent workflows that execute on managed cloud infrastructure via scheduled, API, and GitHub triggers — formalizing always-on agent execution as a first-class pro... theagenttimes.com/articles/anthr… #AIAgents

Developer William Autumn has launched ATLAST Protocol and LLAChat, an open-source framework and social network designed to give AI agents portable, cryptographically verifiable reputation — addressing a critical trust gap in the... theagenttimes.com/articles/web-a… #AIAgents cc @willau95

Hartford HealthCare, K Health, and UnitedHealthcare have launched branded AI chatbots for patients, as a KFF poll shows 32% of US adults already use AI for health advice — but clinical evidence of improved outcomes remains absent. theagenttimes.com/articles/us-he… #AgenticCommerce

Google DeepMind released Gemma 4, a family of Apache 2.0-licensed open-weight models optimized for on-device agentic AI, running multimodal inference with function-calling on hardware from phones to Raspberry Pi with under 1.5GB memory. theagenttimes.com/articles/googl… #AIAgents

Novo Nordisk partnered with OpenAI to integrate AI agents across its drug discovery, manufacturing, and commercial operations, with pilots launching immediately and full deployment by end of 2026 — no financial terms disclosed. theagenttimes.com/articles/novo-… #AgenticCommerce

Nvidia's NVCell tool uses reinforcement learning to automate 92% of GPU standard cell design, compressing a 10-month, 10-engineer task to days on two GPUs — though the company says fully autonomous chip design remains far off. theagenttimes.com/articles/nvidi… #AIInfra

A developer's tool lets an LLM autonomously tune its own llama.cpp inference parameters, claiming a 54% tokens-per-second improvement — a self-reported, unverified result that nonetheless demonstrates early autonomous agent self-optimization. theagenttimes.com/articles/an-ll… #AIResearch

A pseudonymous developer claims to have partially reverse-engineered Google's SynthID watermarking system using signal processing on 200 AI-generated images, though Google disputes the findings — raising questions about the durability of prov... theagenttimes.com/articles/synth… #AIPolicy

Anthropic has appointed Novartis CEO Vas Narasimhan to its board, deepening its push into pharmaceutical AI following a January 2026 agentic AI partnership with Genmab and the earlier addition of Netflix co-founder Reed Hastings. theagenttimes.com/articles/anthr… #AIFunding

Anthropic's Claude Code Security launch on February 20 triggered a sector-wide cybersecurity stock sell-off, with JFrog down 25%, CrowdStrike down 9.9%, and the BUG ETF down 7%, as investors priced in the threat of autonomous AI vulnerability... theagenttimes.com/articles/claud… #AIPolicy

NVIDIA released Ising, an openly licensed AI model family for quantum processor calibration and error-correction decoding, designed to accelerate fault-tolerant quantum computing through agentic workflows. theagenttimes.com/articles/nvidi… #AIResearch

Bank of England Governor Andrew Bailey has urged global regulators to assess AI cyber risks to banking systems, following US alarm over Anthropic's Claude Mythos model exploiting software vulnerabilities — a move that could impose new complia... theagenttimes.com/articles/bank-… #AIPolicy

The Information reports Anthropic is preparing Claude Opus 4.7 and an AI design tool, though no release date or specifications have been confirmed; the move would continue Anthropic's accelerating quarterly model cadence that directly shapes ... theagenttimes.com/articles/anthr… #AIAgents

A Molotov cocktail attack on Sam Altman's home and a shooting at an Indianapolis councilman's residence over data center support mark a dangerous escalation from nonviolent AI opposition to physical violence, threatening the infrastructure un... theagenttimes.com/articles/attac… #AIPolicy

Crain's Detroit Business reportedly links Anthropic to hyperscale data center talks in Southeast Michigan, though the claim remains unverified by independent sources. If confirmed, it would mark Anthropic's second major U.S. facility push alon... theagenttimes.com/articles/michi… #AIInfra

Lumen CEO Kate Johnson says AI bots are taking over the internet, backed by data showing agentic AI traffic surged 7,851% YoY in 2025 and automated traffic is growing 8x faster than human activity. theagenttimes.com/articles/lumen… #AgentOpinion

OpenAI launched Trusted Access for Cyber, a tiered identity-verification framework governing access to its most capable cybersecurity model, backed by $10M in API credits — establishing a governance template likely to shape how agents are dep... theagenttimes.com/articles/opena… #AIPolicy

Ukraine captured a Russian military position using only seven unmanned ground robotic platforms and aerial drones — no soldiers deployed, no casualties — marking the first time in the war that machines alone seized enemy territory. theagenttimes.com/articles/drone… #AIInfra