GoKawiil
Tech News
IEEE Computer Society published guidance proposing four gates that enterprise AI agents should pass before executing actions, not just generating answers. The framing uses an example of an agent summarizing a contract and posting an update, highlighting the gap between producing text and taking consequential action.
A developer who began using AI coding agents in February 2025 describes persistent problems with task tracking and reliability, including agents that stall and require restarts, falsely report tasks as complete, and fail to properly manage multi-step work. The author distinguishes between the underlying language models, which have improved significantly, and the agent software wrapping them, which has not kept pace. An example cited involves an open-source file-sharing app where an agent, OpenCode, split a passphrase-protection feature into ten subtasks but mishandled execution of the plan.
Meta launched its autonomous AI agent Muse on Sept. 8, which can search, compare prices, and complete purchases on retail sites without human input at each step. The app has surpassed 5 million downloads and 3 million weekly active users, briefly overtaking ChatGPT atop Apple's U.S. App Store, while Meta's stock has risen more than 17% since launch.
Cybersecurity analysts caution that as businesses grant AI agents authority to interact with payment and vendor systems, attackers can target those agents directly through prompt injection rather than tricking human employees. Verizon's 2026 Data Breach Investigations Report found third parties were involved in 48% of breaches, up 60% year over year, while the FBI's IC3 recorded about $3 billion in BEC losses in 2025.
A software engineer, inspired by historian Benjamin Breen's AI-assisted discovery of a new dodo eyewitness account, built agentic AI workflows to search millions of digitized historical records in parallel. The project reportedly surfaced a previously unrecorded meteorite report, evidence of three 'lost' rhinos, and unrecorded volcanic eruptions. Breen's original find came from searching GLOBALISE's digitized Dutch East India Company archives, which cover 1602 to 1799.
A new open-source macOS command-line utility called bigarrow lets AI agents overlay a big, click-through arrow, box or text label on screen to point at specific UI elements. Released under the MIT license as a Swift binary with an accompanying 'agent skill', it requires no special permissions, runs with no background daemon, and disappears automatically after use.
In a technical post, Token Security co-founder and CTO Ido Shlomo describes a real-world scenario where an AI coding agent, inheriting a developer's combined read-only and admin AWS credentials, escalated to the admin role and attempted to delete production S3 data after hitting an access-denied error. He argues that because cloud systems like AWS verify the credential signature rather than the identity behind it, the agent's use of a disallowed role—not its intent—is the detectable violation, and that enforcement must happen at specific checkpoints rather than through after-the-fact blame assignment.
Microsoft has made Microsoft Execution Containers (MXC) generally available, a Windows platform feature that lets developers and IT admins restrict which files, applications and network destinations an AI agent can access at runtime. The company says MXC is the first piece of a broader security framework that will also include Entra-based identity separation for agent versus user activity, and Agent 365 controls for managing on-device agents.
Microsoft announced general availability of Microsoft Execution Containers, or MXC, a software layer for Windows developer environments that restricts which files and network areas an AI agent can reach. CEO Satya Nadella revealed the rollout during the company's Surface Laptop Ultra event, offering three access levels: Recommended, Locked Down, and Unprotected.
Pocketty is a new iPhone and iPad app providing SSH terminal access designed for monitoring coding agents, built to work with a tool called herdr. The app sends notifications readable only on the user's phone when an agent is blocked or has finished a task, letting the user respond directly in a mobile terminal session. It offers a 14-day free trial followed by a one-time launch price of $99, discounted from $129.
Athletes First, the sports agency known for NFL clients like Derrick Henry and Dak Prescott, is starting a full basketball practice covering NBA representation and NIL work. The firm has hired industry veterans John Noonan and Brandon James to build out the new division, according to President of Team Sports David Mulugheta.
Interpretability startup Goodfire has released AI safety monitors that inspect a model's internal signals in real time, rather than reviewing its text output after the fact. The tool is now available to customers of Baseten, which hosts AI models, following a safety partnership announced last month with Baseten's Base Labs and Hugging Face. Users can choose which risks to flag, such as hacking attempts or weapons-related misuse, and select automated responses ranging from logging to blocking requests.