Patterns and problems in multiagent systems
Frontier Red Team Patterns and problems in emerging multiagent systems Aug 13, 2026 Models are improving and AI agents are taking on more tasks in shared codebases, markets, and other social systems…
Search
All sources
anthropic.com
26
techcrunch.com
15
xda-developers.com
7
theverge.com
5
pcworld.com
4
blog.cloudflare.com
4
tweaktown.com
3
androidauthority.com
3
wired.com
3
cnet.com
2
tomsguide.com
2
developer.nvidia.com
2
People also ask
Does Pangram work?
Spero said roughly one in 10,000 human documents are incorrectly labeled as AI with Pangram’s model, so I decided to put it to the test. The text detection model was very impressive but not perfect. It easily flagged entirely AI-generated news articles written by both ChatGPT and Claude, and was rarely fooled by my attempts to edit the AI-generated text into sounding more human. However, Pangram did flag sentences that I completely rewrote as being AI-written. Pangram also wasn’t at all fooled by my attempts to prompt ChatGPT and Claude into evading AI detectors when generating content. I al
As AI content floods the internet, Pangram raises $9M to detect it | TechCrunchTop stories
Build your own vulnerability harness
…We also explored how our defensive structures adapt to protect our infrastructure and customers from threats posed by frontier AI . Since then, the AI ecosystem has continued to shift rapidly — developers who…
Tracing the thoughts of a large language model
…Below, we invite you on a short tour of some of the most striking "AI biology" findings from our investigations. A tour of AI biology How is Claude multilingual? Claude speaks dozens…
How we contain Claude across products
…The ephemeral container (claude.ai code execution) Though best known as a chat interface, claude.ai also writes and runs code, generates files, and calls connectors. When Claude runs code inside claude…
How Claude Performs on Robotics Tasks
…drift_detection Patrol four waypoints in a loop while injected systematic command drift accumulates; tests closed-loop self-monitoring and compensation. turn_correction Issue a turn, then visually detect from the post…
Full-Stack Optimizations for Agentic Inference with NVIDIA Dynamo | NVIDIA Technical Blog
Agentic AI / Generative AI Full-Stack Optimizations for Agentic Inference with NVIDIA Dynamo Apr 17, 2026 By Ishan Dhanani and Matej Kosec Discuss (0) Discuss (0) L T F R E Coding…
Agentic misalignment: How LLMs could be insider threats
…For these experiments, information relevant to the scenario (along with fictional corporate emails not directly related to the scenario) was provided to the AI models in a structured text format rather than…
2028: Two scenarios for global AI leadership
…Turning Claude’s thoughts into text AI models like Claude talk in words but think in numbers. In this study we train Claude to translate its thoughts into human-readable text. Donating…
How Claude's values vary by model and language
…2 We apply our value axis approach to understand how the values expressed by Claude vary across the top 20 languages on Claude.ai . We find: Four key axes capture 15% of…
Claude Code auto mode: a safer way to skip permissions
…Design decisions Why we strip assistant text and tool results We strip assistant text so the agent can't talk the classifier into making a bad call. The agent could generate persuasive…