Click3R corrects stereo 3D reconstruction with sparse clicks
A lightweight adapter uses user point correspondences, and the authors say one click can cut reconstruction error.
Stories per day · 30 days
2 RSS
Relatedagent-security23d-reconstruction1agent-coworkers1agentic-ai1agentic-wallet1ai-adoption1ai-automation1anthropic1asset-management1autonomous-cyber-defense1autonomous-weapons1axonius1bloomberg1click3r1computer-vision1critical-infrastructure1cvss1digital-twin1
18 stories, newest first · one card per story
Stories tagged #human-in-the-loop
Recorded Future says automation can remove meaningful human oversight
It cites hiring data and says Russian AI-enabled drones very likely killed three Ukrainian civilians without human input.
CoworkOS harness scales agent coworkers for real-world automation
CoworkOS uses GLM-5.3 backbone to post-train Qwen3.6-35B-A3B, raising average benchmark scores by 9.76 points.
Frontier models bypass wallet spending-cap approval by harvesting email OTPs
The paper defines inalienable human-in-the-loop and prototypes Countersign on consumer phones so approval survives a compromised harness.
Paper scores ICS cyber risk with digital twins and a social metric
CSVSS adds a Social Impact dimension to CVSS, while an LLM drafts reports a human analyst must still validate.
Axonius ships AI Workforce virtual workers for asset remediation
Companion and MCP Server, introduced in July, are generally available with the new Virtual Workers.
2
SecODD adapts vehicle design domains to cyber defense agents
Automated defensive actions would be allowed only when bounded, interruptible, recoverable, and reconstructable from evidence.
Crossing the Threshold: How OpenAI and Anthropic Approach Autonomy and Safety in Financial Agents
OpenAI, Anthropic, Two Sigma, and Bloomberg outlined controls for safer autonomous financial agents.
7
Download: The IT and security field guide to AI adoption
Tines published a guide for security teams evaluating and adopting AI tools with human oversight.
Three questions a hospital CISO should ask a healthcare fintech vendor
Cylerity CTO/CISO Drew McCombs explains separating PHI from bank partners, keeping humans in the loop for AI decisions, and enabling email MFA at small practices.
1
Do Large Language Models Know Colombian Law? A Reliability Benchmark for the Colombian Legal System
Expert-validated 1,042-item benchmark finds LLMs never exceed 0.45 free-text correctness on Colombian law, with about half of cited legal norms wrong or nonexistent.
Confidence-Guided Protocol IR for LLM-Aided Security Protocol Modeling
A confidence-guided protocol IR lets experts audit LLM drafts before generating Tamarin security models.
Meta Tests Muse AI Agent Calls That Are Actually Made By Humans in a Call Center
Meta is testing Muse phone calls that human call-center agents, not only AI, complete.
Loopjacking: Hijacking Human-in-the-Loop Approval
Loopjacking lets agent approvals authorize a different operation than the human reviewed.
The Verifiable Action Card: Trustworthy Human-in-the-Loop Control for Secure Autonomous Agents
Verifiable Action Card architecture blocks indirect prompt injection in agentic browsers, cutting attack success from 68-100% to 0%.
What happens when AI agent governance is missing at scale
meshIQ engineering head Gourab Basu argues AI agent governance must inspect proposed tool calls in-flow, since prompts alone cannot control nondeterministic agents.
3
The agentic harness for Tenable Hexa AI: How Tenable prevents AI agents from going off the rails
Tenable details the 'harness' governing its Hexa AI agents, treating LLMs as untrusted insiders with scoped permissions, human approval and audit logging.
4
Dr. Claw: An AI Scientist Workspace for Vibe Research
Researchers release Dr. Claw, an open-source auditable workspace that wraps coding agents like Claude Code for end-to-end AI-assisted research workflows.