#安全

共 100 篇带此标签

100 articles Newest 2026-11-21
Archive · #安全 11 articles
LLMs

LLM Weight Rewriting Is Not Self-Evolution

A close look at arXiv paper 2611.21885, which explores how language models can modify their own weights—and why evaluation, rollback, and reproducibility matter more than a single benchmark gain.

2026-11-21 Source 2026-11-20 8 min 48 views
Case Studies

Palo Alto Networks' $4.5B Bet on the Agent Runtime

Palo Alto Networks is acquiring Applied Compute for $4.5 billion to bring cloud infrastructure and security controls together for AI Agents. The key questions are pricing, migration, multicloud compatibility, and whether the combined platform can win developer trust.

2026-11-19 Source 2026-11-18 6 min 48 views
Opinion

Code Review Is the New Bottleneck for AI Agents

As AI coding agents accelerate software production, code review becomes the constraint that slows delivery. Engineering teams need redesigned risk routing, smaller changes, independent validation, and clear accountability rather than another review bot.

2026-11-18 Source 2026-11-17 6 min 48 views
Opinion

Windows 11 May Be Tracking Claude Code Usage

Reports suggest Windows 11 may collect usage data from AI coding agents such as Claude Code and Codex. The key concern is not ordinary telemetry, but an operating system gaining visibility into how developers put Agents to work.

2026-11-17 Source 2026-11-16 7 min 48 views
Opinion

Google’s Search Limits for AI-Skeptical Schools

Google reportedly plans to reduce search visibility for schools that explicitly reject or restrict AI. The dispute raises a broader question: when search rankings affect procurement and public visibility, how much institutional autonomy is actually left?

2026-11-17 Source 2026-11-03 6 min 48 views
Opinion

AI Agents Expose the Cost of Technical Debt

The PHP2Shell case shows how AI can surface vulnerabilities faster than organizations can patch them. Closing the gap depends on better SBOMs, tests, deployment records, and release processes.

2026-11-16 Source 2026-11-04 7 min 48 views
Tools

Brave Adds x402: Agents Can Now Buy Search

Brave Search API now accepts machine payments through x402, allowing AI Agents to purchase search calls per task. Developers still need to enforce spending limits, idempotency, and approval controls.

2026-11-16 Source 2026-11-13 7 min 48 views
Opinion

Can Model Weights Carry Viruses? LLM Supply Chain Risks

Malicious patterns may be embedded in model weights and propagate through downloads, quantization, fine-tuning, and redistribution. Although the current evidence remains limited, production teams should pin model sources, record hashes, and audit the exact weights they deploy.

2026-11-15 Source 2026-11-13 7 min 48 views
Case Studies

Can LLM Weights Reveal the GPUs Used for Training?

Unemployment Labs is testing whether statistical traces in LLM checkpoints can identify export-controlled GPUs used during training. The approach could give investigators a model-side signal, but it remains probabilistic rather than courtroom-grade evidence.

2026-11-13 Source 2026-11-12 6 min 48 views
Tools

Pomerado Turns Websites into MCP Tools for AI Agents

Pomerado aims to wrap websites that lack APIs into callable MCP interfaces within minutes. This article examines how it works, how developers connect to it, and the risks of running it in production.

2026-11-12 Source 2026-11-11 7 min 48 views