Neura News

AI News

News reporting focused on AI and machine learning, covering the companies behind these technologies, their real-world applications, and the ethical concerns they raise. This includes areas like generative AI (large language models, text-to-image and video), speech tech, and predictive analytics.

Latest News

8 articles
Technology

Multi-Agent Workflows Beat Single AI Agents for Complex Business Tasks, Forbes Argues

A Forbes article by Bernard Marr argues that single AI agents are insufficient for complex business tasks, advocating for multi-agent workflows where specialized AIs handle distinct parts of a process. The piece provides real-world examples in marketing and customer service, along with a six-step design guide. Marr emphasizes modularity, clear hand-offs, and human intervention points, warning against giving agents too much power.

Aug 246 minNeura News
Research

The Week AI Moved From Monoliths to Networks

Last week, the AI industry shifted from monolithic labs to networked development. Google split its research leadership, with Jeff Dean launching Discovery Loop and Demis Hassabis stepping back at DeepMind. Meta released Muse Code, a coding agent that orchestrates sub-agents. Major infrastructure deals, including Anthropic's $10B compute agreement with Volta and SK hynix's $38B fab investment, underscored the trend. New ventures and research tools also emerged, signaling a move toward orchestrated, multi-agent systems.

Aug 97 minNeura News
Research

New Benchmark Measures How Multi-Agent Systems Fail and Recover

OrchestraBench, a new benchmark introduced in an arXiv paper, uses controlled failure injection to measure how multi-agent systems fail and recover. It introduces metrics like cascade radius and per-failure-mode recovery, revealing that simple routers fail on adversarial cases while intent-reasoning models succeed. The benchmark also identifies three tiers of failure handling and shows that blind retry amplifies latent faults.

Aug 74 minNeura News
Research

PlanFlip Attacks Target Multi-Agent LLM Planning Phase

New research from Yuhang Wang introduces PlanFlip, a framework of four prompt injection attacks targeting the planning phase of multi-agent LLM systems. The attacks exploit a single injection into the Planner agent's context to corrupt all downstream sub-tasks. Testing on nine frontier LLMs across 3,479 episodes revealed that stronger models like GPT-5 are more vulnerable, while reasoning-augmented models like DeepSeek-R1 show full resistance.

Jul 213 minNeura News