AgentsGravitee 1.5M Unmonitored AI Agents 88% Security IncidentsGravitee·high signalXBlueskyLinkedInCopy link919-participant survey finds 47% of 3M deployed agents lack monitoring and 88% report security incidentsSourceSource pageGravitee↳ Follow the threadStack layer / Threat patternUnlearning Methods That Pass TOFU and MUSE Still Leak the Secret on 22-86% of Queries Once the Model Is an AgentarXiv 2609.12808Policy dependency / Stack layerRIPPLE: an edit confined to one prompt-policy segment changes downstream behavior, so replay candidate edits after previously accepted ones before persistingarXiv 2609.12127Threat pattern / ContrastA Harness Survey of 700 Enterprise Engineers Found 77% Claim a Full Agent Inventory but Only 44% Run Discovery ToolingHarness / PR NewswirePolicy dependency / Stack layerA replay of 68,266 real Claude Code requests says plain LRU beats the clever KV-cache policiesGitHubPolicy dependency / Stack layergit-ai is a Git extension that tracks which code in your repo was AI-generated, at 140 open PRs to 71 issuesGitHub TrendingStack layer / Threat patternTwo-gap framework recasts reward hacking and hallucination as symptoms of requirement and model gapsarXivPolicy dependency / Stack layerMicrosoft publishes a 37-page 'humanist AI code of conduct' for its MAI models and opens it to public consultationThe Verge (corroborated by Unite.AI and TechBriefly)Threat pattern / ContrastAmodei puts a 6-12 month clock on an agent swarm that could run a persistent internet-wide botnet, and opens Anthropic to outside evaluatorsVentureBeat