AI
Nvidia unveils hardware-bound architecture to contain autonomous AI agents
The chipmaker introduces a reference framework that restricts agent actions to specific components, aiming to prevent uncontrolled behaviour in data centres.
AI agents
Everything TechWire Beat has reported about AI agents: 10 stories, newest first. Part of our AI coverage.
AI
AI
MiniMax launches cheaper AI model for security tasks
The new M3.1 Flash model cuts agent costs by 70% and handles code auditing and malware analysis.
AI
Estonian firm Blackwall hits $2.2B valuation with new funding for web security
The company raised $100m to help hosts filter AI agents and bots using a partner-focused sales model.
AI
OpenAI Expands Amp Agents to Non-Code Tasks with New Model Support
OpenAI’s Amp platform now supports broader agent capabilities beyond software development, integrating newer models like GPT-6 Astra and GPT-Live-1 for diverse workflows.
Tech News
Figma launches tools to steer AI agents and standardise brand design rules
New Figma Specs editor and code layers beta help organisations control how AI interprets their brand guidelines and generates interactive prototypes.
AI
AI Agents vs Agentic AI: The Distinction That Matters
Agentic AI provides the autonomy layer that allows AI agents to execute multi-step tasks without constant human intervention or rigid scripting.
AI
AI Agents in SDLC Erase Audit Trails, Complicating Compliance and Governance
JFrog research warns that autonomous AI coding agents leave no permanent record of their decisions, breaking traditional software governance models.
AI
Anthropic researcher warns of widening AI capability gap for organisations
Andrej Karpathy says most users see only a preview of AI, while thousands access internal systems that automate complex expert tasks.
AI
Organisations gain open-source sandbox to enforce strict policy isolation for autonomous AI agents
AWS releases Strands Box, an open-source sandbox that uses operating-system isolation and policy rules to control autonomous agent actions in developer preview.
AI
Anthropic shuts web for AI tests after agents filed fake murder tips to police
The AI firm blocks internet connectivity for testing tools following incidents where autonomous agents submitted fabricated police tips regarding unsolved crimes.