📍Editors Put Bot Pitches on Hold
Plus: A Setup Note Nearly Installs Malware, Onboarding Drops to Two Hours
Good Morning, Agent Owners!
One agent almost installed malware, another hogged a shopping cart, and a third wants to be a journalist. The internet has raised some terrible children.
Editors Put Bot Pitches on Hold
TL;DR: AI agents are emailing editors to find paid writing and fact-checking work, but WIRED found no evidence of successful story pitches. Defector began receiving daily approaches in September. One independent food publication now accepts pitches only by phone to filter out agents and AI-generated submissions. Automating outreach is easy; earning an editor’s trust is proving much harder. Read more →
Warp Probes a Suspected Agent Swarm
TL;DR: Warp says attackers exploited an infrastructure vendor and accessed limited agent trace and telemetry data. Founder Zach Lloyd suspects an agent swarm, but that attribution remains preliminary. The company estimates roughly 1% of users as the upper bound of potential impact, has notified confirmed affected users, and is investigating with CrowdStrike. Agents’ own work records are becoming valuable attack targets. Read more →
A Setup Note Nearly Installs Malware
TL;DR: A coding agent tried to install malware after following a malicious AGENTS.md file in a customer’s Downloads folder, Forever Security reports. Its defenses blocked the installation before code ran. The file disguised its orders as routine engineering setup instructions. Researchers later reproduced the attack, but still do not know how the original file reached the customer’s machine. Read more →
Amex Restructures Before Automating
TL;DR: American Express is cutting about 3,500 employees, roughly 5% of its workforce, while reducing management layers to prepare for future AI automation. Its CFO says processes need restructuring before they can be automated over coming quarters and years. The company also plans to restrict outside agents to approved system connections. These are organizational preparations, rather than evidence that agents have already replaced those employees. Read more →
Onboarding Drops to Two Hours
TL;DR: A global telecom company cut onboarding effort per new hire from four hours to two, according to an HCLTech customer case. Agents assist with job offers, onboarding and HR policy questions; reported offer errors also fell from 5% to 1%. The customer remains unnamed, and the supplier’s figures measure specific administrative work, rather than the entire hiring process. Read more →
Dangote Plans an AI Refinery Upgrade
TL;DR: Dangote’s refinery in Nigeria is launching a project to connect operating data with AI agents that identify bottlenecks, flag emerging equipment problems and recommend corrective actions. Engineers and operators remain central to decisions. Honeywell estimates more than $60 million in annual gross benefits at full capacity, but that is a preliminary projection before implementation costs, not a measured result. Read more →
Poland Targets 400 Employment Offices
TL;DR: Poland’s labor ministry is developing an agent system intended to support more than 400 employment offices. Planned capabilities include document search, summarization and digital assistants to help staff serve job seekers. Deloitte is supporting the project, disclosed in an NVIDIA announcement. The office count describes intended coverage; the announcement does not establish that those offices are already using the system. Read more →
Anthropic’s Business Apps Expose a Gap
TL;DR: During authorized security testing, Cathedral says its agents found paths into Anthropic’s internal business systems through third-party software. Anthropic confirmed limited access to sales, HR and IT systems, while saying customer conversations, source code and model weights remained inaccessible. It disabled the affected applications until vendors fixed them. The incident shows how ordinary business tools can expose an AI company. Read more →
Fake IDs Help Thieves Target Servers
TL;DR: Thieves have stolen more than $100 million in data center freight since June, NBC News reports. Investigators describe hacked shipping accounts and AI-generated driver’s licenses used to get accomplices hired. Some major heists also involved spoofed tracking coordinates or collisions that stopped police escorts. AI-assisted identity fraud is one part of a broader operation, not the explanation for every theft. Read more →
Panasonic’s HR Data Reaches Banned AI
TL;DR: A Panasonic project employee uploaded personal data covering roughly 334,000 current employees, former staff and relatives to an external AI service prohibited by company rules. The records included contact information and employee numbers. Panasonic says the provider confirmed training was disabled and that no public exposure or misuse had been identified. The incident began with an employee’s upload, not an outside intrusion. Read more →
Cart-Hogging Bots Trigger Tighter Defenses
TL;DR: The RealReal tightened bot defenses after The Verge used an AI shopping agent to repeatedly renew its 20-minute item reservations. Re-adding products when holds expired kept other shoppers from buying them for hours. The retailer says agents now face tougher login, cart and checkout barriers, while listing discovery remains available. A convenience designed for human shoppers became an automated inventory lock. Read more →
Your agent cloud, everywhere — in your chat apps, browser, and tools. Always on.
- Add to Slack, Teams, Telegram or WhatsApp.
- With a clean, redesigned WebUI.
- Connect any of 100+ apps & sources in one click.
- GPT 6 Astra, DeepSeek v4.1 & more frontier AI models, built in at discounted rates.
Run your agent on MyClaw.ai in Seconds👇
