Nobel Peace Prize laureate Maria Ressa describes AI's takeover in three phases: social media cloned you and microtargeted your attention; LLM chatbots moved into your intimacy; and now agentic AI — systems you give permission to act for you — puts something deeper at stake. Her framing is argument, not incident report. But the practical question it raises is real and answerable: who currently has permission to act as you?
Key figure: 63% of organizations in IBM's 2025 breach study had no AI governance policy, and 97% of AI-breach victims had no proper AI access controls (IBM, Cost of a Data Breach Report 2025).
Short answer: every "Sign in with Google" click, every shopping or scheduling agent, every API key and OAuth grant is a piece of your agency you've delegated. Most people have dozens they forgot. Audit them tonight: revoke what you don't recognize, narrow what you keep, and put approval steps on anything that can spend money.
Answer honestly.
Google: myaccount.google.com → Security → Third-party access. Microsoft: account.microsoft.com → Privacy → Apps and services. Apple: Settings → your name → Sign in with Apple. Remove anything you don't recognize or no longer use. Full Gmail/Drive access for a coupon app is a revoke.
Shopping agents, scheduling assistants, coding agents, "auto-pilot" tools — list every agent that can act without asking you first. If one was compromised or went rogue, what could it do right now? That's your blast radius.
API keys, "full access" integrations, email plugins that can read and send. Narrow scopes to the minimum the tool needs; rotate keys you haven't touched in months.
Check what your most-used AI chatbot stores about you and whether conversations train future models. Delete or opt out where the option exists.
Subscriptions, saved payment methods, one-click buying, agents with checkout permission. Set spending caps and require approval above a threshold you choose.
Source: Foreign Policy reel (Sep 29, 2026), full interview at foreignpolicy.com/2026/09/25/maria-ressa-big-tech-democracy-ai-autocracy/ — quotes from the reel's own transcript.
Built by Ingenious AI — commentary evaluated on its own merits; Ressa’s framing presented as her argument. Quotes from the reel’s transcript. This page is general information, not legal advice.
A review of every AI agent, app, and integration that holds permission to act on your behalf — what each can do, and whether it still should.
Shopping and scheduling agents, ‘sign in with’ OAuth grants, API keys, and AI assistants with tool access — most people hold dozens they have forgotten.
Remove unused OAuth grants and API keys, narrow broad permissions, and put approval steps on anything that can spend money or send messages.
In her September 2026 Foreign Policy interview, Ressa described three phases: social media's microtargeting took our attention, LLM chatbots took our intimacy, and now agentic AI — systems you give permission to act for you — puts our agency itself at stake. It is her argument and framing, presented as commentary.
Google: myaccount.google.com > Security > Third-party access. Microsoft: account.microsoft.com > Privacy > Apps and services. Apple: Settings > your name > Sign in with Apple / iCloud connected apps. Remove anything you don't recognize.
A review of everything authorized to act in your name: OAuth app grants (Google, Microsoft, Apple), AI agents with standing permissions, broad API access to your email/files, and anything that can spend money automatically. Revoke what you don't recognize or no longer use.
Filed under AI Agents 2026 by Ingenious AI — see the related checks above for more in this series.