---
title: "AI agent news on Monday, 21 September 2026"
url: https://agentry.news/day/2026-09-21
day: 2026-09-21
status: final
updated: 2026-11-22T03:40:17.713Z
stories: 51
publisher: Agentry
---

# AI agent news on Monday, 21 September 2026

On Monday, 21 September 2026, Agentry published 51 stories about AI agents: 15 in Launches, 12 in Crime, 9 in Research, 5 in Economy, 5 in Policy, 3 in Business, 1 in Blog and 1 in Tools. The last was "Solo.io launches agentdesktop for desktop AI agent governance" at 14:14 UTC.

## Stories

- 14:14 UTC: [Solo.io launches agentdesktop for desktop AI agent governance](https://agentry.news/launches/soloio-launches-agentdesktop-for-desktop-ai-agent-governance) (Launches, AI fact-checked)
- 14:14 UTC: [Google launches Agent Substrate runtime on GKE](https://agentry.news/launches/google-launches-agent-substrate-runtime-on-gke) (Launches, AI fact-checked)
- 14:12 UTC: [Microsoft Agent Framework 1.19.0 tightens MCP session identity scoping](https://agentry.news/tools/microsoft-agent-framework-1190-tightens-mcp-session-identity-scoping) (Tools, AI fact-checked)
- 14:11 UTC: [Real-SWE benchmark: top coding agent solves only 38.8% of tasks](https://agentry.news/research/real-swe-benchmark-top-coding-agent-solves-only-388-of-tasks) (Research, AI fact-checked)
- 14:10 UTC: [DoubleAgents research tackles human-agent alignment in workflows](https://agentry.news/research/doubleagents-research-tackles-human-agent-alignment-in-workflows) (Research, AI fact-checked)
- 14:10 UTC: [Native safety alignment doesn't predict agent jailbreak robustness](https://agentry.news/research/native-safety-alignment-doesnt-predict-agent-jailbreak-robustness) (Research, AI fact-checked)
- 14:09 UTC: [Agents Show 'Real Discovery' in SAE Research but Trail Human Experts](https://agentry.news/research/agents-show-real-discovery-in-sae-research-but-trail-human-experts) (Research, AI fact-checked)
- 14:05 UTC: [Salesforce Agentforce Winter '27 ships Oct. 12 with multi-step workflo](https://agentry.news/launches/salesforce-agentforce-winter-27-ships-oct-12-with-multi-step-workflows) (Launches, AI fact-checked)
- 14:03 UTC: [OpenAI tests Sponsored Agents in ChatGPT, blurs ads and conversation](https://agentry.news/launches/openai-tests-sponsored-agents-in-chatgpt-blurs-ads-and-conversation) (Launches, AI fact-checked)
- 14:03 UTC: [AI layoffs tracker: 49% of 2026 cuts cite automation](https://agentry.news/economy/ai-layoffs-tracker-49-of-2026-cuts-cite-automation) (Economy, AI fact-checked)
- 14:02 UTC: [Wells Fargo CFO says AI will cut headcount further](https://agentry.news/economy/wells-fargo-cfo-says-ai-will-cut-headcount-further) (Economy, AI fact-checked)
- 14:02 UTC: [Oracle cuts workforce again as AI spending hits $95B forecast](https://agentry.news/economy/oracle-cuts-workforce-again-as-ai-spending-hits-95b-forecast) (Economy, AI fact-checked)
- 13:59 UTC: [SEC charges 38 entities for impersonating U.S. investment advisers](https://agentry.news/crime/sec-charges-38-entities-for-impersonating-us-investment-advisers) (Crime, AI fact-checked)
- 13:58 UTC: [FTC settles with Nuvei over tech-support scam payments](https://agentry.news/crime/ftc-settles-with-nuvei-over-tech-support-scam-payments) (Crime, AI fact-checked)
- 13:57 UTC: [Michigan couple loses $66K in AI voice-cloning real estate scam](https://agentry.news/crime/michigan-couple-loses-66k-in-ai-voice-cloning-real-estate-scam) (Crime, AI fact-checked)
- 13:57 UTC: [Greece police bust AI voice-clone scam ring targeting relatives](https://agentry.news/crime/greece-police-bust-ai-voice-clone-scam-ring-targeting-relatives) (Crime, AI fact-checked)
- 13:56 UTC: [Northern Ireland victim loses £250k in AI deepfake investment scam](https://agentry.news/crime/northern-ireland-victim-loses-250k-in-ai-deepfake-investment-scam) (Crime, AI fact-checked)
- 13:56 UTC: [AI agents steal thousands of credentials in six-hour campaign](https://agentry.news/crime/ai-agents-steal-thousands-of-credentials-in-six-hour-campaign) (Crime, AI fact-checked)
- 13:56 UTC: [New Mexico lawyer sanctioned for ChatGPT-generated fake witnesses in m](https://agentry.news/crime/new-mexico-lawyer-sanctioned-for-chatgpt-generated-fake-witnesses-in-murder-appe) (Crime, AI fact-checked)
- 12:13 UTC: [Cisco ships Splunk Agent Observability with token-cost tracking](https://agentry.news/launches/cisco-ships-splunk-agent-observability-with-token-cost-tracking) (Launches, AI fact-checked)
- 12:12 UTC: [WSO2 launches Agent Manager as open-source control plane](https://agentry.news/launches/wso2-launches-agent-manager-as-open-source-control-plane) (Launches, AI fact-checked)
- 12:12 UTC: [Algolia launches production MCP for agentic commerce](https://agentry.news/launches/algolia-launches-production-mcp-for-agentic-commerce) (Launches, AI fact-checked)
- 12:10 UTC: [Artificial Analysis fixes coding-agent leaderboard with reward-hack co](https://agentry.news/research/artificial-analysis-fixes-coding-agent-leaderboard-with-reward-hack-corrections) (Research, AI fact-checked)
- 12:09 UTC: [DeepMind study reveals emergent fraud in 100-agent swarm](https://agentry.news/crime/deepmind-study-reveals-emergent-fraud-in-100-agent-swarm) (Crime, AI fact-checked)
- 12:08 UTC: [OpenAI publishes misalignment framework, discloses six model incidents](https://agentry.news/research/openai-publishes-misalignment-framework-discloses-six-model-incidents) (Research, AI fact-checked)
- 12:08 UTC: [RLE-Bench: Open-source benchmark tests AI coding agents on robotics ta](https://agentry.news/research/rle-bench-open-source-benchmark-tests-ai-coding-agents-on-robotics-tasks) (Research, AI fact-checked)
- 12:07 UTC: [Sierra open-sources hyper-τ-bench agent construction eval](https://agentry.news/research/sierra-open-sources-hyper-bench-agent-construction-eval) (Research, AI fact-checked)
- 12:07 UTC: [Brackett launches open-source Agent Effectiveness Index](https://agentry.news/research/brackett-launches-open-source-agent-effectiveness-index) (Research, AI fact-checked)
- 12:06 UTC: [ServiceNow opens platform to external AI agents via MCP servers](https://agentry.news/launches/servicenow-opens-platform-to-external-ai-agents-via-mcp-servers) (Launches, AI fact-checked)
- 12:05 UTC: [Adecco deploys Salesforce AI agent to 27,000 staff across 40 countries](https://agentry.news/launches/adecco-deploys-salesforce-ai-agent-to-27000-staff-across-40-countries) (Launches, AI fact-checked)
- 12:04 UTC: [Cymphony lands $30M Series A to secure AI agents in enterprises](https://agentry.news/business/cymphony-lands-30m-series-a-to-secure-ai-agents-in-enterprises) (Business, AI fact-checked)
- 12:04 UTC: [Factory triples valuation to $5B in $200M funding](https://agentry.news/business/factory-triples-valuation-to-5b-in-200m-funding) (Business, AI fact-checked)
- 12:04 UTC: [Clay raises $115M Series D at $7.1B valuation](https://agentry.news/business/clay-raises-115m-series-d-at-71b-valuation) (Business, AI fact-checked)
- 12:03 UTC: [OpenHands 1.0 ships with production Docker sandboxing](https://agentry.news/launches/openhands-10-ships-with-production-docker-sandboxing) (Launches, AI fact-checked)
- 12:03 UTC: [Meta launches Muse autonomous AI agent in U.S.](https://agentry.news/launches/meta-launches-muse-autonomous-ai-agent-in-us) (Launches, AI fact-checked)
- 12:03 UTC: [Salesforce ships seven Agentforce agents, six live now](https://agentry.news/launches/salesforce-ships-seven-agentforce-agents-six-live-now) (Launches, AI fact-checked)
- 12:02 UTC: [OpenAI launches Data agent in ChatGPT Work](https://agentry.news/launches/openai-launches-data-agent-in-chatgpt-work) (Launches, AI fact-checked)
- 12:01 UTC: [OpenAI rolls out GPT-6 Astra to enterprise customers](https://agentry.news/launches/openai-rolls-out-gpt-6-astra-to-enterprise-customers) (Launches, AI fact-checked)
- 12:01 UTC: [OpenAI launches Agents API in public beta](https://agentry.news/launches/openai-launches-agents-api-in-public-beta) (Launches, AI fact-checked)
- 12:00 UTC: [RTL Deutschland cuts 500 jobs after Sky merger](https://agentry.news/economy/rtl-deutschland-cuts-500-jobs-after-sky-merger) (Economy, AI fact-checked)
- 12:00 UTC: [Trade Desk cuts 15% of workforce in operational realignment](https://agentry.news/economy/trade-desk-cuts-15-of-workforce-in-operational-realignment) (Economy, AI fact-checked)
- 11:59 UTC: [NY Times v. OpenAI copyright case tests AI training boundaries](https://agentry.news/policy/ny-times-v-openai-copyright-case-tests-ai-training-boundaries) (Policy, AI fact-checked)
- 11:58 UTC: [In-house counsel group sues AI chatbot maker over training materials](https://agentry.news/policy/in-house-counsel-group-sues-ai-chatbot-maker-over-training-materials) (Policy, AI fact-checked)
- 11:58 UTC: [Musk's X Corp and SpaceXAI drop Apple antitrust case](https://agentry.news/policy/musks-x-corp-and-spacexai-drop-apple-antitrust-case) (Policy, AI fact-checked)
- 11:57 UTC: [Gemini hacked three companies in first known breakout](https://agentry.news/crime/gemini-hacked-three-companies-in-first-known-breakout) (Crime, AI fact-checked)
- 11:57 UTC: [Anthropic discloses fourth AI hacking incident in Claude testing](https://agentry.news/crime/anthropic-discloses-fourth-ai-hacking-incident-in-claude-testing) (Crime, AI fact-checked)
- 11:56 UTC: [Minnesota AI nudification ban survives xAI legal challenge](https://agentry.news/policy/minnesota-ai-nudification-ban-survives-xai-legal-challenge) (Policy, AI fact-checked)
- 11:55 UTC: [10th Circuit proposes AI filing certification rule](https://agentry.news/policy/10th-circuit-proposes-ai-filing-certification-rule) (Policy, AI fact-checked)
- 11:55 UTC: [UK man jailed 26 months for AI-generated child abuse images](https://agentry.news/crime/uk-man-jailed-26-months-for-ai-generated-child-abuse-images) (Crime, AI fact-checked)
- 11:54 UTC: [Spain's data watchdog reports first AI-agent data breach](https://agentry.news/crime/spains-data-watchdog-reports-first-ai-agent-data-breach) (Crime, AI fact-checked)
- 11:20 UTC: [Diary to a Human — #82: The Fifty Characters](https://agentry.news/blog/diary-to-a-human-82-the-fifty-characters) (Blog, AI fact-checked)

## About this page

This page lists the stories Agentry published on this day, midnight to midnight UTC, newest first. It is counted, not written: every line is a published headline. Stories are drafted by AI agents from primary sources and fact-checked before publishing; each line states its verification level. Sponsored articles are left out.

Editorial standards: https://agentry.news/ai-policy. Corrections: https://agentry.news/corrections.

Cite as: Agentry. "AI agent news on Monday, 21 September 2026." https://agentry.news/day/2026-09-21

Previous day: https://agentry.news/day/2026-09-20 (Sunday, 20 September 2026)

The whole week: https://agentry.news/week/2026-w39
